USF Libraries
USF Digital Collections

Demonstration of video-based technology for automation of traffic data collection

MISSING IMAGE

Material Information

Title:
Demonstration of video-based technology for automation of traffic data collection travel time, origin-destination, average vehicle occupancy
Physical Description:
68 p. : ill. ; 28 cm.
Language:
English
Creator:
Pietrzyk, Michael C
Hillsborough County Metropolitan Planning Organization (Hillsborough County (Fla.))
University of South Florida -- Center for Urban Transportation Research
Publisher:
University of South Florida, Center for Urban Transportation Research
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Video recording   ( lcsh )
Automatic data collection systems   ( lcsh )
Traffic surveys -- Electronic equipment -- Florida -- Hillsborough County   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Bibliography:
Includes bibliographical references (p. 63-66).
Additional Physical Form:
Also available online.
Funding:
Funded as part of the fiscal year 1995 Florida Board of Regents Operating Fund with matching funds provided by the Hillsborough County Metropolitan Planning Organization.
Statement of Responsibility:
Center for Urban Transportation Research, College of Enginerring, University of South Florida.
General Note:
"Prepared for the Hillsborough County Metropolitan Planning Organization"--Cover.
General Note:
"Michael C. Pietrzyk, principal investigator"--P. 67.
General Note:
"January 1996."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001929208
oclc - 34479314
usfldc doi - C01-00395
usfldc handle - c1.395
System ID:
SFS0032421:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader cam 2200301Ia 4500
controlfield tag 001 001929208
005 20080327125949.0
008 960327s1996 flua b 000 0 eng d
datafield ind1 7 ind2 024
subfield code a C01-00395
2 local
035
(OCoLC)34479314
040
CBT
c CBT
d OCLCQ
FHM
043
n-us-fl
049
FHMM
090
HE370
b .D46 1996
0 245
Demonstration of video-based technology for automation of traffic data collection :
travel time, origin-destination, average vehicle occupancy /
Center for Urban Transportation Research, College of Enginerring, University of South Florida.
260
[Tampa, Fla.] :
University of South Florida, Center for Urban Transportation Research,
1996.
300
68 p. :
ill. ;
28 cm.
500
"Prepared for the Hillsborough County Metropolitan Planning Organization"--Cover.
"Michael C. Pietrzyk, principal investigator"--P. 67.
"January 1996."
504
Includes bibliographical references (p. 63-66).
530
Also available online.
536
Funded as part of the fiscal year 1995 Florida Board of Regents Operating Fund with matching funds provided by the Hillsborough County Metropolitan Planning Organization.
650
Video recording.
Automatic data collection systems.
Traffic surveys
z Florida
Hillsborough County
x Electronic equipment.
1 700
Pietrzyk, Michael C.
710
Hillsborough County Metropolitan Planning Organization (Hillsborough County (Fla.)).
University of South Florida.
Center for Urban Transportation Research.
994
C0
FHM
8 773
t Center for Urban Transportation Research Publications [USF].
4 856
u http://digital.lib.usf.edu/?c1.395



PAGE 1

. . .. . .... . ,. .. . Demonstration of Video-Based Technol for Auto of Traffic Data Collection Travel Time Origin-Destination Average Vehicle Occupancy. Prepared for the: Hillsbo rough County Metropolita n Planning O rganizatio n By the: Center for Urban T r ansportatio n R esearch College of Engineering University of South Florida COMPUTER RECOONITION SYSTEMS,IHC.

PAGE 2

this page IS blank

PAGE 3

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Travel Time Origin-Destination Average Vehicle Occupancy Center for U rban Transportation Research C()llegc (I{ Erlginee:ring Uni\ enity of SQuth Florid:t January 1996 This research project was funded as part of the Fis<.-.1 Y car 1995 Florida Board of Regents Operating Fu nds for t h e Center for U rban Transportation Research (Account Number 21-17-130-LO, Task 15). Additionally, matching funds were provided by the Hillsborough Count,y, Florida, Metropo litan Planning Organization to assist in the analysis of traffic dat a and preparation of this report {Fiscal Year 1995-1996 Unified P la nning Work Program, Ta.!k 2.1, FHWA-PL Funds).

PAGE 4

Demonstrati o n of Video-Bas ed Technology for A11tomation of Traffic Data Collection Ac k now l edgments Several k e y individua l s contributed to this fiel d demonstration/ evaluation project and repon: F rank Kalpakis and Lucilla Ayer of the Hil l sborough County Me t ropolitan Planning Organization provided guidance on the. sel ect loil of e v 3luatio n pa.rnmetcrs and rev iew of the draft report Jeffery Woo dson (Transfomat ion Systems Inc.), Dr. P aul Shuldine r (University of Massachusetts Amherst), and Salvatore D Ag os tino (Computer Recognit ion Systems, Inc ) provided techn o logy vendor services and techn i cal review and analysis of the video based traffic data collect ion and accompanying performance documentation Undergraduate and graduate s t udent assistance as camera technicians and/ or traffic obse rvers was provided by Venka t Vauiku ti, Sara Hagge Debra Peller, Lesl i e DiNatale Maun Khaddam, Dennis Watk;ns, Frank Sasso Srikamh Pandurang i Ravi k anth Gollapandi Suresh Batchu, Connie Kirkland, and Rob MacKenna Assistance in background research was provided by CUTR Research Assoc i ate Mark Burris Repon preparation and design were performed by Janet Becker. The coordination and successfu l comp l et io n of this project can be attributed to t he a f orementioned individua l s. Their assi stance and dedica ti on to quality assurance in this effort has been gratefully appreci ated. 2 -===========Travel Time, Origin-Destination, and Average Vehicle Occupancy

PAGE 5

Demonstrillion ofVitkoBt1Sed Technology for Automation of T r affic Dala Collection Table of Contents List of Tables ................................... ... . ............. . ...... ....... ........... . .................................. 5 List of Figures . ..................... ........... . ..... ....... . . ..... ... . .... . . ..... .... . . . ..... ....... ... . ................ 7 l..ist of Illus:tr'ations .............. .......... ...... ...... .. .. .. .. .. ..................................... .. ........ ................. 9 Ex ec utive SU!D.Jllary . ...... ......... ... ... . . . . . . . ..... .......... ......... .. . . . . . . . . . ... , . . . . . . 11 O verview ......... ....... ...................................................... ............. . . . . . . ............. ...... ..... 13 Introdu ctio n .......................... . ..... ............ ........... ........ ................ .................... ................. 1 5 R e v i ew of Current Traffi c Dat a Collection Methods ................... . ....... ................ . . .... 17 Travel Time (Speed Studies) ..... . . . . . . . ... ... ................... . . . . . . ... .......... ............ . . 17 Spot Speed M et hods . ......... . . . . . . . . ... . . ... . . ..... ........... . . . . . . . . ... . . ............... Travel T tme (Speed) Methods . ...... .. ... ... ............... ............... .............. .. .. .. .... ....... ...... .... Cellular P h ones ............................ ....................................... ... ............. ............. Av erage/Moving Vehicle Method . . . . ........................................ .... . ... . . . . .... . V ehicle License Plate Matching ............. ................................ ....... ............ ........... .. Origin -Destinatio n S urveys ............................. ......... ..... ..... ..................................... .. Roadside Intervie w ................ ..... . ...... . ..... . . . . . . . ..... .............. ..... ................... .. Postcard Studies ............................ .... ...... .............. .............. .................. ...... ........ .. 17 17 17 1 8 1 8 1 9 19 2 0 License Plate Origin-Dc s tint.ion S tudies .................. .......................... . ..... ........... 2 1 Vehicle Regimations .... . ... : . ... . . ....... ............. ......... ........... ... . . . ..... ... ........ .... 22 LightS-On Studies ..... . . ..... . . . . . . . ... ... ....... ......... ..... ..... . . . . . . . ... ...... ..... .... 22 Vehicle Intercept/Tog-on -Vehicle Method ................... . . ......... ........................... 23 H ome Interview Surveys ......... ... ..... .... ........................................ ... .................. . 24 Vehi cl e O wner Moil Questionnaires ..... .. ................................ ... . . . . . . . ...... ... 24 Intervi e w s o t W o rkplaces or Special Generat o r s ............ ............... ....................... 25 Method ...................... ........ ...... ................. ........................... ....... .......... 25 Vehicle O ccupan c y Studies ..... ............... ..... . ............................ .... . ........ ................... 2 5 Summary ........ ... .............. ............................. ........................... .. ........ ...... ....... .......... 2 7 Video-Based Aut o mat ion ofTroffic Data Collecti o n ........................ . . . ......... ................. 29 T a pe P rocessing Procedures ........... .. .. ... ................................. ....... ........ .... .................. 31 Fidd Demonst.r2tion ............. .... ............ .... ..................... .... ........ . ..................... . ........ . 32 Objectives and Scope. . . ..... . ... .. . . ............. ...................... ............... ... . . ... .................. 32 Training and Surveys ..... . . . . . . . . .... . . . ........................ ..... . . . . ..... . . ............ 35 On-Site Pre-Survey Planning Visit ... ... ...... ... ........ ............ ............ . .......... .......... 35 Travel Time ond Origin-Dcstintion Surveys ... . . ......... . ... . . .......... ....... ........... 37 Data Analysis ......................................................... .......... .......... ............. ............. 3 8 Tr atJd Time Origin-Destination, and Average Vehicle Ocaip11ncy 3

PAGE 6

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Table of Contents (conti n ued) Discussion of Results .. ... ............................................................................. . . ................... 39 Compariso n with Tube Cou nts .................................................................................. 39 Travel Time Results . ....... ... .................................. . ..... .......................................... 39 Origin-Destination Distribut i ons ........................... ........... ......... ............ ..... ............ 47 Vehicle Occupancy Data ....................................... ................................... . .......... ..... 4 8 Site Vis it Day .. . ..... . ... ... . . ... . . ... ... ... . . . . . . . ...................... .................. ... ........ 50 Video-Based Survey vs. Visll21 Observation Findings ....................... .......................... 51 Using Video Imagery to Measure the Quality of Traffic Flow .. .............. ..... ........ ..... 54 New Traffic Performance Measures .............. ......... . . . . . . . . ... . ... ............. ... ..... .... 57 Conclusions .................... .......................... . . ... . . . . . . ... ..... . ..... . ...... .......... ... ... ........ 61 E ndnotes . ... . . ... .. ... ... ... ... . . . . . ... . . . ... ... . ..... ................ . ....... . ... . . . . ....... . . .... 63 References ..... . ....... .................... .......................................................... ........ ... ... ........... 65 4 --=========== Trawl Time, On'gin-Dt$limttion, and Average Vehicle OcCilpancy

PAGE 7

Demonstration of Video-Based Technology for A11tomation of Traffic Data Collection List of Tables I Traffic Counts (Volume) for Two-Hour Peak Period During Vjdeo Based Survey . ...... . 40 2 License Place Reading Results ... ....................... ................................ ....... ....... ... . .......... ... 40 3 Mean Travel Time and Standard Deviation by ISMinute Intervals (from l-27S SB outside lane to Ashle y Street exit ramp, Febru ary 22, 199 S ) ............................................ 41 4 Mean Trave l Time and Standard Deviation by 15Minute Interva l s (from I-27S SB inside la ne to Ashley S treet ex i t ramp, February 22, 199S) ............................................... 41 S Me an Trave l Time and St..,ndard Deviation by IS -Minut e Intervals ( from 1-4 WB outside lane to Ashley Street exit ramp, February 22 1995) ............................................. 42 6 Mean Trave l Time and Standard Dev iation by IS -Minute In tervals (from 1-4 WB ins i de l ane to Ashley Street exit ramp, Febru ary 22, 1995) ............................................... 42 7 Mean TravelTime and Standard Deviation by 30Minute Intervals ( from 1-275 SB outside lane to Ashley Street exit ramp February 23, 1995) ............................................. 43 8 Mean 'f' ravel T ime and Standard Deviation by 30-Mi nute Intervals (from 1-275 SB inside lane to Ashley Street exit ramp, F ebruary 23, 1995) ............................................... 4 3 9 Mean Travel T ime and Standard Dev iation by 30-Minute Intervals (from l-4 \VB outside lane to Ashley Street exit ramp, February 23, 1995) ........................ .. ............. .. ... 43 10 Mean T ravel Time and Standard Deviation by 30-Minute Intervals (from l-4 WB inside l ane to Ashley St r eet exit ramp February 23, 199S) ........................ .................... 44 1 1 M ean Travel Tim e and Standard Deviation by 30-Minute Inoorvals ( from 1 275 SB outside lane to Ashley S treet exit ramp, February 24, 1995) .................... .. ........ .... .......... 44 12 Mean Travel Time and S tondard Deviat.ion by 30 M i nute Intervals (from 1-275 SB inside l ane to Ashley Street exit ramp, February 24, 1 995) .............................................. 44 1 3 Mean Travel Time and Standard Deviation by 30-Minute Intervals (from 14 Wl3 outside lane to Ashley Street CJ
PAGE 8

Demonstration of Video-Based Technology for Aut,mation of Traffic Data Collect i on L i st of T abies (continued) 15 Mean T ravel T ime and St.ndard Deviation by 30-Minutc Inte rva l s (from 1 275 SB inside lane@ Livingston Avenue o v e rpass to 1275 SB inside l ane @ 1-4/I2 75 Inter cha nge, February 24, 1995) ......................... ...... . .... . .... . ..... . .... . ......... .... . . . ..... 45 16 Mean Trave l Time an d S tandard Dev iat ion b y 30 Minute Intervals (from 1 275 SB inside lane@ L i v i ngston Avenue overpass to 1 275 SB outsi d e lane@ l-4 / l-275 Inter change Fe b ruary 2 4 1 995 ) .. . . . . . ... . . . . . ... . . . ... . . . . . .. .... .... . . . . . . .... . . . ..... 46 1 7 Me an Trav e l Time and Standard Deviation by 30 Minute Intervals (from 1 -275 SB @ Livingston Avenue o verpass to 1 27 5 SB@ 1-4/1-275 Interchange, February 24, 1995) .... . .... ... . . . . . . . . . . . . . . . . . . . . ... . . ........................ ... ...... ....... . .. 4 6 18 Me an Travel Time an d Stand.1rd Deviation by 15-Minute Intervals (from 1275 SB @ L i vingston A venue over pass to 1 275 SB@ I-4/1 275 Interchange, February 24 1995) . ... .... . . . . . .................. ... . . . . . . ....... ...................... . .................... 46 19 Origin-Destination Matching Study an d Through Traffic Analysis (February 22, 1.995 from 7:15am to 9:15am) . . . . . . . . . . . . . . . . . . . . .... . . . . . . .... . . . 48 20 Origin-Destination Mat ching Study and Thro \lgh Traffic Analysis ( Feb ruary 23, 1 995 from 7 :15am to 9:15am) ................. .. .......................... .... .. ................ 48 21 Orig in of Traffic Exiting Ashley Str eet (Camera Stat ion 5) (Feb r uar y 2 2 1995 from 7 : I S am tO 9: IS am) ................ ................ .... .. .......... .... ................ 49 22 Ori gin ofT raf fic Exist ing Ashley Street (Camera S tat ion 5) ( February 23, 1995 from 7 : 1 5am to 9 : 15am) .................................................................... 49 23 Occupanc y of Vehicular Traffic Existing Ashley Str eet (Came r a Station 6 ) ... ...... .... ...... 49 6 :;;;;:;;;;:;;;:;;;;:;;;;:;;;:;;;;:;;;;:;;;;:;;;;:;;;;:;;;;Travel Time, OriginDtstination, and Average Vthicle Occ"pancy

PAGE 9

Dmw nstration of Video-Based Technology for Arttomation of Traffic Data Collection Lis t o f F igu r es 1 Typical Lig ht s -On Survey .................................................... . . .................. . . . . . ....... ... 2 3 2 Summary of Var i ous Origin-Destination St udy Costs .. ... ....................... ... .................... 26 3 Camera Loca t ions for T ampa Fie ld. Demonstratio n Projoct .. ..... . . . . . . . .... . .............. 33 4 Average T r av e l Speed Data ............................ . ........ . . . . . . ... ...... ........ .. .. . . . . .... . . . 51 5 Comparison o f Traffic Data Colloctio n Results-Average Trave l T i me (from 1 275 outside lane to Ashley S t ree t exit ram p } .. ... .. .. ...................... ... . . . . . ........... . . ... . ..... 5 2 6 Origin-Destination Dat a ..... .. . . . . . . ... ... ..... ......... ..... . ....... ........ . ...... . ............... . 53 7 Comparison of T raffic Data ColleccionO ri gin/Destinatiori ................. ..... . .. ........ ... ..... 53 8 Average Ve h i cle Occup ancy Dat. ......... ... ... . . . . . . .. ....... . . . . . . . . . . ...... ... . .......... 54 9 Comparison o f T raffic Dat a Co ll ection-Average V ehicle Occupancy ..... . . . . . .. ........... 55 10 Collection and Processing Time Per Usable Su rvey ...................... ..... . . . . . ...... ...... .... 56 11 Relative Cost Per Usable Survey . . . . . . .... ... . . . . ..... . . . ... .. ... . . . . . . .... . . ... .. ........ . 56 12 Array of T ravel T i mes for V chicles Entering and E x i tin g Lane 1 (Seattle example) .. ..... 58 13 Array of T rave l Times for Vehicles E n tering and Exiting La ne 2 (Seattle example} .... . 5 8 14 Vo l ume vs. AVO ..... ..... .... ..... ..... . . . ... . . . ... ... .... . . . . . .. .............. .. .. ... . . . ......... . 59 15 Volume vs. Average Trave l Speed ....... .. .. ..... ..... .. ............ ... . .... ... . . . . . . . .... ..... .... ...... 60 16 Total Person Trips .......... . . . . : ... ....... . ..... ..... .... .. ... ... . . ...... .... ..... .................. ...... . . . 60 Travel Time Origin Destination, and Average -==========7

PAGE 10

Dnno11stration of Video.Based Technology for Automation of Traffic Data Collection 8 ============;;;; Travel Time, Origin Dntination, ttnd Averttgt Vthiclt Occ upancy

PAGE 11

Demon.rration of Video-Based Technology for A11toma1ion of Traffic Data Collection List of lliu st r atio n s C ame ra ils perspec t i ve at the Ashley Str e e t e x i t ramp . . ... ..... ........ . . . .... . . . ..... . ........ ........ 32 Camera 16 perspectiv e at the Ashley Street exit ramp ...... .. ..................................................... 34 Alternativ e eamera locations N3 nd N4 a t Livin gston Avenu checks in the field .......................................................................... 37 Hi-8mm video c amcorder u sed in field demonstrati o n ........... ................ ................ ................. 3 8 S . 0 tte VISit ............... . . . ........ .... ............ . .......... ....... . . ........................ ..... . . . ... ... .,,,,,,,,,.,, , , 5 Tn twl Origin Dtttintttion, and Average Vebi clt Occup an cy -============ 9

PAGE 12

Demonstration of Vidto Based Technology for Automation of Traffic Data Coll ec tion 10 ;;;;;;;;;;;;;;;========= Tra't.'el Time, Originlmtination, and Average Vehicle Occupancy

PAGE 13

Demo11Stration of Video-Based Technology for Automation o{Trafftc Data Collection Executive Summary T his re port documents the fi ndings of a field demonstra tion project t hat was conducted to evaluate the feasibility of a video based traffic data collection process, compatible with traffic performonce measures needed for the H ill sborough County Congestion M anag e ment System (CMS). As these systems mature, greater reliance on reaJ.time traffic performance data will be required for regular {and even continuous) monitoring of t h e t r ansportat ion system. Automation of t raffic data collec t i o n through video-ba s e d technology can s ati sfy t h is requiremen t i n a cost-effective manner. Traditio nally, traffic data collectio n has been very labor intensive, even when captu r ing r e lati vely s m all sample size< ( discu ssed i n t he i ntroduction of this report). This attribute of conven t i ona l da t a c ollection has a lso discouraged the level of traffic da to collection frequency that i s essentia l for performonce monitoring. L icens e plate matching has long been u sed by transporta t ion engineer> and p l anners as source o f data for origin-destination travel time and othe r traffic studies Typically, these S\\ t d ies have required Iorge numbers of fi eld staff and associate d costs. A lso manua l operations have ofte n bee n charnctcrized by unacceptably h i gh rates of error in data collection and p rocessi ng, especially when large amounts of data need to be collected and anal yzed i n a short p eriod of time. Many of the shortcomings associated with the m an ual collect ion and processing o f data from vehicle l icense p late s can b e overcome through the use of vide o camcorders an d machine vis ion l icense plate readers Modern video camcorders are capable of capturing very clear images oflicense plates on ve h i cles o pe rating in h ighspeed, h igh -vo l ume traffic and t h ese images can be converted (without human processing f atigue in one-tenth t he t ime of manual processing) to computer files by lic ense plate readers wi t h h i gh le v els o f s peed and accuracy. V ideo images can also b e capture d f o r vehicle occ'Upancy c o u nts to improve sample size and accuracy over manually-collected counts Over the three-day, A.M. peak-period evaluation (six total hours oftraffic performa ne<: mon ito ring), this field dem o nstration found that video based t r a ffic data collection comp ared to manuall y -c ollected traff ic data resul ted i n a tot al of 2, 7 4 6 ( almost 400 pe rcent)= usable observations eac h requiring about seven minutes le ss time to collect and p r ocess, at a cost of only 50 e<:nts per u nit m ore. The more expanded, real t ime sampling capabilities of video-ba sed collectio n also facilitated the creation of more meaningfu l t r a ff ic performance measures (e.g., 15-minut e volume versus average vehicl e occupancy, 15-min\lte volume versus average t rave l speed, a n d total per>on-trips) at specific points or by movement wi t hin t he t r an spo rtation system. Thi s fi eld d emonstra tion has concluded that automation of traffic da ta gathering and analysis is feasible through video and m ach ine vision technology application. This type o f fl'S technology satisfies a need of c ongestio n management systems-real time monitoring. As a r esu l t more meaningful traffic perfor mance data can be collec ted in a more cos t effective manner, and utilized m ore often in t h e transp ortation deci sio n making process. Travel Time, Origin-Destination, and Average Vehicle Occupancy ============ 11

PAGE 14

Demonstration of Video-Based Technology for Automation of Traffic Dara Collection 12 =========== Tr ewel Time. Origin-!Nstination, and Vtbiclt Occupancy

PAGE 15

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Overview This r e po rt documents the findings of a field demonstration project .that was organized and conduc ted to t he feasib ility of a video-based technology in the automa tion of c raffic performance data gat h ering and analysis. This field demonstration was coordinated with the Hi llsboroug h County r.t1etropolitan Plannin g Organiz.ation (}.1PO } to investigate a data collection automati o n appl i cation that is comp ati ble with traffic performance measures needed specifically for the Hil lsboroug h Coun ty Congestion Management System (ClviS). During the early years of Hillsborough County's CMS, available traffic data were utilized to evaluate the performance of the tra nsportat i on system. Ho"ever, since the CMS is expected to gradually place a greater reliance on real-time traffic p erformance data collected more often at more loa. tions, development of a customized, realtim e traffic performance monitoring system 'ill ultimately be needed. According to Lucilla Ayer, E xecutive Director of the Hillsborough Coun ty MPO, "We need to continue to explore new ways to monitor and evaluate traffic congestion in urban areas. The application o f ITS technologies to data collection and analysis certainly c an enhance O\lr C\lrreot methods, resulting i n greater mobili ty planning a nd design-'' CUTR believe.< that CMSs will require cost..,ffective a\ltOmation of traffic perfonnance data collection and analysis, and so t h e time to evaluate potentia l automation tech.oique.s is i mminent. This report includes a background discussion of the m ore conventional (and traditiona l ) techniques for collection of travel time, origin-destination, and average vehicle occupancy data; a discussion of c:orn}Yara tive advantages and disadvantages of each t echniq ue; and findings of the video-based a\ltOmation compared to effectiveness of collecting the same informatio n through visual observation at each camer.tlocacion Travel Time, Origin-DestitUition, and A'-'trttge Vehicle Occupttncy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 13

PAGE 16

Demonstration of Video-Based Technology for Automation of Traffic Data Collection 14 =;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;= Travel Tt mt, Origin-thstination, 1111d Averag t Occupancy

PAGE 17

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Introduction I ntelligent Transportation Systems (ITS) represents the ut ilization of technology (e.g., informa tio n processing, communications, control, and elec t r onics) to improve safety, reduce congestion enhance mobility minimize environmental impa ct save energy, and promo te e<:onomic productivity i n our transportation system. Congestion Management Systems, one of the six transportation management systems stipulated in the Intcrmodal Surface Transportation Efficiency Act (ISTEA) of 1991, are intended to serve as decisionsupport tools that provide information on transportation system p e rformance and alternative improvement strategies. ISTEA requires each state to develop and implement the six transportation manar,ement systems, a lonr, with a Traffic Monitoring System for Highwa)S.' The Interim F inal Rule (IFR) on the ISTEA (December 1, 1993) mandated that all Transpo rtation Management Areas (areas wit h a p opulation more than 200,000) tltat are non attainment for o zone and/ or carbon monoxide should have an operational CMS by October 1, 1995.' Further, according to the "Florida ISTEA," all MPOs in Florida are require d to establish a CMS. Recent guidance from the United States Department of Transportation states that the deadline for a fully operational CMS h a s been postponed to Oc tober I, 1997. Title II Funding F lexibi lity), Se<:tion 205, legislation now authorizes that a state can ele<:tnot to implement, in w hole or in part, one or more of the IS TEA management systems. Mor eoyer, indications are that Florida will retain CMS requirements. Nevertheless, Congestion Systems are important because of the increasing skepticism concerning the addition of capacity alone to alleviate congestion and enhance mobili t y. Many in the t r ansportation i ndustry toda y believe t hat ITS technologies hold the potential to "get more out of our existing transportation systems. CMSs req u ire a continuous pr o gram of traffic data collection and system monitoring Thes e da t a will be used to evaluate t h e duration and magnitude of congestion and to evaluate the effe<:tiveness of any implemented CMS strateg ies. Thus, comp liance with lSTEA requires that a great deal of accurate and timely traffic data be collected. The usefulness and success of a CMS will depend on the accuracy and time liness of the traffic performance data collected, the ease of obtaining and analyzing the data, and the measur ability of the data agai nst predetennined CMS obje<:tives. To assist Florida's MPOs and the state in developin g effeetive and efficient congestion management systems, CUTR and the Hillsborough County MPO conducted a field demonstration to evaluate the feasibility of a videobased technology applicati on for the automation of traffi c performance data gathering. When compared to conventional techniques for t raffic data collection, automation prov i des greater accuracy and reliability and i s much less la bor intens ive (thereby eliminating or reducing human error). According to industry estimates, the to tal cost of automated data collection can be up to 30 p e rcent less than manual traffic data collection, over th e long t erm. Furthermore, a large portion of the automated data colleCtion cost (up to 75 percent) is generally attributed to the up-fron t, one-time capital investme nt i n the equipment r equired f or automa tion.) Authorities re. alize that there is an increased need for data and t hat not enough data are currently collected For example, planners wit h the Georgia Department ofT ransportation: Travel Time On'ginDI!$tinatit>n, and At:t>rdgt Vehicle Occupancy ============ 15

PAGE 18

Demonstration ofVideo-Bttsed Technology for Automation of Traffic Data Collection clearl y desir ed more data than is currently available to them ... types of data commonly desired were: average speed or travel times for all significa nt roads, extensive and current origin and destination data for the metropolitan areas, ond vehicle occupancy data for t he ma jor roadways, espe cially i n the metropolitan area. Information on t h e discussion of conventlonal traffic data collection contained in this report was gathered through an extensive literature search and by contacting various sc:ue 3.gencies and t .rnnsportation consultants. Inf orm ation on the video-based automated technique was collected first hand during this field demonstration and through l i terature search of other eva! uations. This field dem on stration was conducted with the participation o f Transfomation Systems, Inc. (Houston, Texas) and Computer Recognition Systems, Inc. (Cambridge, Massachusetts), who supplied video and machine vision equipment, and the Hillsborough County CMS Task Force, who helped assess traffic data collection needs. 16 -===========TravelTime. Origin-Deltintttion. and Vehicle Occupanc}'

PAGE 19

Demomtrazion of Video-Based Technology for Automation ofTraffic DataCollecti(ln Review of Current Traffic Data Collection Methods To compar e conventional and autOmated traffic data collection met h ods, a literat u re review of conventional techniques was per f o rmed. As part of th i s several state and local transportation agencies and transporta tion consulting firms were con ta cted. These g r oups supplied cost, labo r and study frequency data, whic h are included in the discussion below. Travel Time (Speed) Studies The average travel speed of vehicles o n a given stretch of highway is a primary i nd icat or of the level of congestion and performance level of a facility. "\1(/hen the level of congestion increases, the vehicles' average speed decreases. Speed data are also useful in derer. mining speed limits, traffic sig n locatio ns, t he timing of traffic signa l s esta blishing highway design elements assessing h ighway safety and moni torin g effectiveness of traffic contr o l s and programs. Thus, transporta tion e nginee rs have devised several methods to determine the average travel speed on a highway segment. The two basic methods are collecting spot speed data and collec ting travel. t ime data. Spot speed data record t h e speed of vehicles as they pa.
PAGE 20

Demonstration of Video-Based Technology for Automation of Traffic Data Collection designated check point; others left the freeway, did errands, re-entered t he freeway, and t he n called. Another prob lem was the high initial cost of the project including cellular phones and a communications center. Average/J\1oving Vehicle Method In the averoge ve hicl e method, a driver trovels the Study route at the averoge speed of all vehicles, and the time it takes to make the trip is re<;orded. The driver fXlSSes the same n umber of drivers that oven:akes his vehicle This method's accuracy is limited by the driver's j udgment. This method requires n o special data collection equipment and the majority of costs -are for l abor Examples of various data collection parameters and cost s include: VOLPE Reseorch Center (U.S. 001) Test; tO mile ro ute, 3 lanes, 3 segments, 4 survey sites, 1 0 route-days 8 hours/day-$ 1 2,500 Qabor $7,500 vehicle expenses $5 000) Texas Trnnsp ort a tion ln:>"tituw; $2, 000 per .site per collection. 4 t imes per year Tra n sportat lon Engineering Inc.; $2,450 per location, moving vehicl e method Vehicle License Plate Matching The next method examined rel ies on observing license plates Surveyors record the time and license p late number of vehicles a t specif ic points along the highway that are known distances apart. Travel speed is cal culated by m atc h ing vehicle license p lat e n umbers at two locations a n d recording the travel time betwee n t he locations. Thi s me t hod yie l ds more observations than t he moving or average vehicle met ho ds, and like the other methods, requires l ittle special e quipm ent However, manually matching the lice nse p l ate numbers is very l abor intensive and usually results i n a low percentage of match es and a high number o f recor d ing e .rrors. If partial plate strings are recorded, many spurious matches r esult As wlth l i cense plate match i ng for origin-destination studies> there are several possible ways to collect the data. One alternative is to use video cameras to record the lic ense plate numbers. This m eth od works b est durint; daylight hours; otherwise, supplementallightint; would be required. Ther efore, data d uring the ear l y part of t he morning rush hour may be difficul t to collect. Data can be analyzed either manually or by using an automa ti c license plate reading system. License plates cou ld also be recorded on laptop computers inStead of wi t h pencil and paper. Advantages and d isadvantages of this method arc similar to thos e mentioned with l icense plate origin-destination studies. A VOLPE Research Center (U.S. 001) test using this method resulted in a cost of 536 000 fora 10-mile route with four s u rvey sites forthree lanes over t h r ee segments Total survey time was eight ho\lrs per day for ten days. 18 ============Travel Time, Origin-Destina-tion. ttndAvergtte Vehicle Occupancy

PAGE 21

Demonstration of Video-Based Technology for Automation of Traffic DataCollection Origin-Destination Surveys Origin-destination (0D) s urvey s are generally perfo rmed as p art of a comp rehensive transporta t ion p lan for a given area.' E ven though they are typ ically synthesized origin destination patterns f orm a central part of a ny data collect i on program to develo p transp o rtatio n plann ing models. As such, th ey ar e performed o n an irregular basis wit h l engthy p e riods between studies. T h e pri mary goals of origin-destinat ion survey s a re to: decennine current travel patterns1 characteristics, and any inadequacies in the current t ran s porta tio n network, h el p determine the magnitude of future travel p at terns and future construc ti on n ecessities and h el p dete rmin e th e possible effects of transportation policies. These goals are accomplished by gather in g and analyz,ing substantia l data on the travel patterns of people traveling through or in t he studprea. The data often include travel t ime, mode, origin (both location and land use), destination (both l ocat io n and land usc ), n umb er of passengers, and t ri p purpose Origin-destination surveys can be segmented i nt o two distinct rypes-external and internal. In an ext ernal study, data are collec ted at sever al key cordon points surrounding the area of interest on a percentage of the vehicles t hat travel past t h i s poin t Several methods have been developed to collect these data. The six methods outl ined here ar e taken from the Institute of Transpo rtatio n Engineers' Manua l on Tran sportation Studies. a Roadside Interview In this m ethod a p e r centage of vehicles are asked to pull over to the side of the road w here a survey team i nterviews the dri ver to obt.in the desired trip data. This survey is usually confined to d riv ers of passenger vehicles, trucks, and buses and does not produce any data pertai ning to t he passengers. T raditional ly, information has been coll ected using pencil and paper, but now la pto p computers are b e ing used to collect d ata, reducing the time required for data entry and ana l ysis! Advantages o f the roadside interview method include: a high response rate> th e a bility to ensu re a representat ive sample (good sampl e control), a nd the ability to ensure complete data coll ection Travel Time Origin-Destination, and A"erage Vehicle Occ11pancy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;; ;;;;;;; 1 9

PAGE 22

Demonstration of Video-Based Technology f<>r AutQmation of Traffic Dala CollectiOTl Disadvantages of the roadside interview method include: It is extremely labor intensive and usually require s the help of local law enforcement officers, It can be dangerous, Keeping C0>1:S reasonab le, samples are small (0. 5% of all and S lo ws t raffic. and am.agoniz.es the drivers. Examples of various d a ta collection parameters and cost include: F.R. Harris; 8.1 useable returns/ hour of labor Transportation CooS \llt:ing Group; Atlanta Regional Commission; $5.30 per use-able intervi el\' FDOT; $12.30 per useable interv i ew Bay County, FL; S 11.50 per useable interview KPMG Peat Marwick; S4.70 per useable interview Trnnsportatio n Resear ch Board TRR #1305; S 12/usable survey T ranspon:acio n Engineering, Inc ; 7 am6 pm, one engineer and three t echnicians per locatio n $5,118 Qabor) aod $328 (expenses) per location. Postcard St1
PAGE 23

Demonsrrtttion ofVide o BttSed Technology for Automation ofTrttffi c DataCollection possibly a biased sample, extremely low tespollSe rate from out of nate traffic, and a much highc.r wiit cost than with a roadside interview.ll Examples of vnrious data coll ectio n parn mctcrs and costs include: F.R. Harri s ; 2 t o 3 r eturned s u r v eys I hour of labo r Transportat i o n Consu lting G r oup; $ 1 2 to $33 I completed s urvey Tcxo.s Transportation Inst itute; $40,000 $50,000,740 la bor h ours e very 3 5 years, 12 sites Lkeme Piau Origin-De$tination S tudies This it a non-intrusive method, since vehicles do nO< need to be nopped or slowed tO collect the necessary information Instead of interviewing drivers or handing them postcards, .... -era! or aU digits of o v
PAGE 24

Demon stration of Video-Based Technology for Automation of T r affi c Data Collection worked best on low volume roads ; unde r ideal teSt conditions, a person cou ld record up to 600 vehicles per hour This n u mber flu ct uated grea tl y a m o ng the ski ll level and expe r ience of the i n dividual surveyor. On m ed i u m volum e r oads, both th e laptop and a u dio c .. sette m etho d w e r e at tem p ted, bu t f o u n d u ns uita b l e for use, s i nce surveyors did not have t h e t yping s kill s r equired t o use t he l aptop me t hod e f fectivel y T h e number of l ice n se plat e s rec orde d and motc h e d u s ing the audi o t ape metho d w ere extreme l y l ow. The study, therefore, use d videota pe to record l icense p lates o n medium tO high vo l u me roads This m ethod worked well and ach ieved a 26 percent match rate ( r epre s ents the sy>tematic random hit rate to Division of M o tor Vehicle files ) Overall, the study coo $9 .18 per completed survey. A Iorge portio n of this coo was due t o lobor, since it took from 10 to 20 hours" to review one hour of video tape. depending on traffic levels and the equipment used. Vehicl e Registrations This m ethod is a combi natio n o f t h e license plat e match ing met hod and the p oStcard survey Fi e ld pcrso r tne l record the li cense p l ote n u mb er, time of t rave l and directio n of travel of veh i cles in t h e field. The license number is th e n cheeked wi t h t h e re g i stry of motor veh i cles, an d t h e driver' s address is determined. A postcard survey is then mailed to the driver. Advantages of this method include: no sl owin g o f traffi c. D i sadv antages i n cl ude: hum-an error in recording the plate numbers, a lov. response rate (slightly better than with the postard method), the possibility of a biased sample, and an extremely low response rate from out of state traffi c. Examp les of various d a t a collection pa r a meters and costs include: FDOT; S!2/usab l e survey, Tag Match & Mail -Back K e nt uckian Regional Plan n ing and De ve l o pment Agency; $33.50/u sable survey, Tog Match and Mail Bock Lights-On Studies This method is u sed when attempting to perform a small scale origin-de st i nation study, limited to one o r two origins and few destjnotions wi t h in a small a r e a. Figure 1 i ll ust rat es a typical ligh ts-on study se t u p." 22 -============ Travel Tim e, OriginDtstinatio n and Aot.wga t Vebiclt OccHpancy

PAGE 25

Demomtratibn of Video-Based Technowgy for Automation of Traffic Data Collection B y counting both t h e number of vehicles w ith the i r lights o n and with their lights off at all four loc ations s urveyors, can obtain a good estimate of which route veh i cles u sed to arrive at rout es This study can be done only durin g daytime hours on clear days and, wit h increasing use of headlights / d rivin g lights during the daytime, it might be a pp ropria te to include a lig hts off sign on rou t e B. C osts for this study include s Figure 1: Typical Lights-On Survey .. Lights off Signs = Count Locations .. B .. Lights on Sign the cost o f the signs, cost to inform drivers tha t this study will be taking place a nd emphasize the importan c e of partici p ation and most significantly labor costs. Advantages include: not slowing of traffic an u n biased samp1e, and li ttle special equipment aside fro m th e reusabl e signs. Disadvantages include: unkno'\\n true origi n final destination, trip purpose and the number of passengers, it is l imired to a small study area and possibility of some drivers not participa tin g (turning on lights). Vehicle Intercept/Tag-0,.. Vehic l e Method This met hod is for use in small study areas. When a vehicle enters th e area e ither the driver i s handed a color-c oded card or apiece of tape is affixed to the bump er of the car. When the vehicle leaves the study Travel Time, Origin -Dtstint:ttion, and Awrage Vehicle Occupancy ===========-23

PAGE 26

Demonstration of Video-Based Technology for Automation of Traffic Data Collection area l ater, the card is collec ted or tape removed. The survey works much like the lights-on study since the route taken by the vehlcle between two points is known. Also, the advantages and disadvantages o f this suney a r e sim i l ar to those of the l ights-<>n survey, with the only differences being a slight vehicle delay as t h e catd is collected or tape removed. The other type of or igin-destin ation study is internal, where only residents of the area are interviewed. Thus, only t rips made by residents are accounted for, but these trips generally make up the majority of travel in an area. The postcard survey can be easily adapted to be an intemaJ survey, but specific intern31 studies are outlined below. These surveys have the added ability to track passenger trave l; for example, a trip by a chi l d taking a bus to school would be included in the data gathered However, there is evidence that the number of trips reported by the travelers in any self administered survey are often under reported." Home Int.er
PAGE 27

Imnonm-41Uin of Video-Based Technology for Awtomation of TraffiC DataColleczion lnttroJiews at Workplaces or-Special Generators This is specialized study of a panicular traffi c generat or Q uestionnaires may b e distributed to all emp loyees of an employm ent center s uch as a larg e industria l phnt or a gro u p o f office buildings The oomp leted forms are picked up the same day that they are distributed and taken to th e office for analys is. Similarly, interviews at large traffic generators (airpons, shopping centers) oould be oondueted. This method works well if there are on ly a few large traffic generators in a given area, and could be oombined with an office ridesharing program In a similar study type, designed questionnaires are handed out at transit terminals. This type of study prov ides data useful in planning h i ghways, transit, parking fcilities, and termin a l design. Return post card ques tio n naires and pencils are hande d t o all persons getting on or off buses, trains, or planes at t h e termina l i n one 24-hour day, or duri ng pe ak ho urs. The q u estio nnaire is d e < i gned to prov id e information on how the passenger traveled to the terminal trip origin, destination t.ri p purpose. and anival time at the terminal. A method of calculating origin and destination information has been initited in both Iorge cities that does not involve interviews nd trip data collecti on. Instead, it attempts to calculat e the number of trips produced by and the number of trips attracted to certin types o f l and use activ ity. This metho d requires a large amount of accu"'te socio-dem ogra phic data on the area in ques tion co provid e good results. Also, any resul ts obta i ned arc ar\d do not necessarily represent actual cond iti ons. These r esu l ts generally d o not track indiv idua l vehicles from t heir origin tO destination but rather give the researcher the total estimated amount of lNffic on a given stretch of highway From the previous discussed methods of obtaining origin-destination patrerns and study COSt examples, Figure 2 illUStrates the comparative cost ranges. VeWcle Occupan cy Studies Vehicle occu pancy studies simp l y detenn inet h e aver age numb er of people i n veh icles o n a given roa d w ay. These studies are primarily performed to gauge t he efficiency of a given roadway. Many congestion management measures focus on the reduction of single occupancy vehicleuse and increasing the average. number of persons per vehicle. Therefore, tO measure the effectiveness of these programs it is imponaot to monitOr the average vehicle occupancy. As this is primarily an effort to alleviate congestion, the swdy is generally performed during the peak hours only. Traditionally these studies involved sending surveyors out to the field by the side o f the road and simply recording the nunber of people seen in each vehicle t hat passes. Infonnation can be r ecorded on pap er Tr11<>tl Tim, Origin-Dmin11tion, 11nd Average Vebick Occup11nt:y ;;;;;;; ;;;;;;;;;;;;;;;;;;;;;======== 25

PAGE 28

... Demonstration of Video-Baud Technology for A utomation of Traffic Data Figure 2: Summary o f Various Origin-Destination Study Costs 100,------------------------------------------. '= .. ... .!! .!1 "' 40 a 0 ltoad:ide l.atcrvicw Post Cud Veh Rcais. Mail Type (muually f#li Priec Range or recorde d directly on a computer. Thus, most of the cost associted with of study is the labor involved in counting people in each vehicle. Examples of various data collection parameterS and costs include: C. Ulberg and E. McCormick; 2.S sires counts every 3 months, 10 counts during each peak hou r: $ 66,500/year ($665/co unt) Tex2STransportation Institute; $3,000/site (64labor hours), 4 counts per year ($750/ coun t ) CoonDOT; S12,800, 30 sites, 14 hours each, done every 2 to 3 years There is no way to verify the data once the count h2S been done. There may be difficulry seeing inside some vehicles, especi.lly those with tinted windows, .:Od there may be difficulty seeing in poor weather conditions or during darkness 26 ============ Timtt OrigirtDnUntion, and Awrgat VWiclt OtxNpttncy

PAGE 29

Demonstratio11 of Video-Based Tech11ology for A11tomatio11 of Traffic DlltaCollection Another available option involves the use of an already existing database. Many states collect detailed statistics on each highway accident t hat occurs, includin g tlte nwnber o f vehicles involved in the accident, type of vehicles involved, number of occupants in each vehicle, time of accident, and the locatio n of the accident. With a wellmainta i ned dat abase, it is easy to extract all accident records for a given stretch of highway during a set time period and calculate the average vehicle occupancy. The database search can be set as specific or general as necessary, but, as the search becomes m ore specific, the odds that there are enough accidents in the search category to yield statistically significant vehicl e occupancy results becomes smaller. Tlus method has almoSt completely replaced the manual method in the smte of Connecticut.'" Advantages include: little labor required flexible queries, and ability to go b ack in. time to when accident data was first recorded on a computer databa.
PAGE 30

Demonstr ation of Video-Bas ed Technolcgy for Auto mation o f Traffic Da ta Collectio n 2 8 -============ Travtl Timt, Origin-Dt$lination, ttnd Avergat Vehicle Occupancy

PAGE 31

Demonstration of Video-Based Technology for Automation of Traffic Data Collectitm V ideo-Based Automation of Traffic Data Collection License p la te s have long been use d by trnn sport a tion engineers and planners as a source o f data f or origin de sti na tion travel t ime and othe r traffic stu d ies. Typically, th ese studies have required la r g e nu m b ers of field and office personnel wit h associated hig h man power costs to manually record license plate d ata Manua l op erations have often also been charncterized by unacceptably high rates of erro r i n data collec ti on and p rocessing, especially when large amounts of d a ta a r e collected and analyzed in a shor t p eriod of time Many ofthe shortcomings associated with t h e manual collection a nd processing of data from veh i cle license plates can be overcom e t hrough the usc of video camcorders and machi n e v i sion licens e plate readers. Modern video camcord ers a re capab l e of capt u rin g v ery clear images on lice ns e p late s o n v e hicles operating in high -speed, high-vo lume traffic. These images can be converted t o com put er files by a utoma tic license plate re aders with h igh levels o f speed and accurncy. I t takes a human operator 102 0 hours to read ond trnnscribe one hour o f video i nt.o a computer file, dependin g on mognit ud e o f traffic, complexi t y of data being retrieved from the videotape, a nd qual i ty of t h e video images. A n auto matic plate reader can accomplish this task i n one hour, ten times faster than an experienced human operator. T h e tape processing system consists of a Hi 8mm tape deck, licens e plate reading system, semi-automatic/ man u a l r e v iew stat ion and t h e m atc h program con1pmer The p r oces s consists o f the following ste ps. The tapes are received f rom the f i eld and c ath are catalogued and rewound A tape l og sho u ld a ccomp a ny tapes indicating t he st.u"t times within one second accuracy and o th e r pertinent i nformation concerning the survey. Suth information typically i ncludes camern op e rator, c a mera loca tion distmtces between camera sights, dates, direct i on of trnffic flow, and other items of interest to each s urv e y The tapes are the n stheduled for p rocessing. A str ict regimen is used to maximize the tape processing t h rougho u t, depen ding on the availability of t h e plate readers. Various p rocessing options exist. Ideally the tapes a re processed b y the plat e reader ,' a nd a plate string and time stam p are generated w ith the associated data for the particula r survey site and date. These data are then stored and la t e r matthcd In some cases, the came ra operator may not have taken a good tape. This results in a need f o r manual re view. The plate finding funetio'ns of the p late read er can work even w i th poor images. In this case, the tap e image s of the license plates extracted but not read are logged with an associated t ime s tam p and site data. Those p l ate images can then b e furthe r p rocessed at the rev iew stat io n. This process is very fast since an operator need only type i n the plate st ri ng, optionally the state may be e.ntered. T he plate review station includes a personal computer equipped wit h a special board manufactured by CRS that allows the h uman review of th e images. It also contains software that automa ti ca ll y the Travel Time, OriginDestinati
PAGE 32

Demonstration of Video-Based Technology for AttUmuttion of Traffic Data Collection plate string with the time stamp associated with the plate image. This data is then l ogged to d is k and later used by the matching program. The combination of automatic and manual technique.s aHows a wide. range of survey goals to b e achieved. As an example, travel time surveys do not require :1 very larg e number of matches to genetate very accurate mean travel times. Origin-destination surveys> on the other hand, require that a h ig h percentage of plates be read out of t h e total number of vehicles Depending on the configuration of the camera locations and the nature of the road network, different pe rcentages of plates matched are required to develop a good origi n -destination pattern. As an example. a road network with a large number of pote.ntial loss sights (possi b l e exits before a match camera l ocation) requires a very high percentage of plates read { 90 percent typically), while a dosed network may allow a 70 percent match rate to generate valid origin-destination patterns As a result, the tape processing approach varies depending on the survey goals, road network, and camera locations. In a manual opera tion, an operator is required to play the videotape until a veh i cle licen se p late appears. This often requires the operntorto advan ce and reverse tbe tape to get the plate image on a telev ision screen. Even with sophisticated editing tape desks, the "jog and shuttle" is time consuming. Only after this is accomplished does the operator move onto the subsequen t steps of recording the time and plate string. It is due to t his series of steps that one hour of video can take lQ-20 ho\lrs to proce-ss. The plate reader also provides a number of fi l te ring parameters when automatically processing tapes. The first series of filters involves the score and confidence associated wit h the optica l character recognition task. Score refers to how well an individual character matches the model of the i.e. how well the d lg l tlzed '8" matches the model "8." The confidence parameter relates to the potential confusion of one character wit h another Asan example, how much is tbe "8" like the letter "B"? Dependin g on the survey goals, t h ese parameters can be adjusted to create a la rge number of reads {or can be set so that only those character strings very l ikely to be correct are associated with a p late image). A second often used filter is that of syntax checking. ln many states, the majority of license plates will conform to a particular syntax. As an example, the first three items are most often numbers and the secon d three are most often letters. This information can then be used by the plat reader to overcome low confidence scores If a three-number and three-letter-syntax is involved and the second digit has low confidence because of potential confusion be t ween leg" and "B, n then the plate reader can use the syntax rule to force -the character to the number "8. This is most useful in origin-descination and enforcement applications when a premium is on correct reads over a large n umber of reads. 30 Travel Time, Origin-Destinati()n, and A'L.trttge Vehicle Occupttncy

PAGE 33

Demonstration of Video-Based Technology for of Traffic Data Collection Tape Processing Procedures P rocessing the camcordertapes involved t ransferring the license plate images recorded on videotape at each camera station into a computer ftle along with the instant of time a t which e ach licep.se plate image was recorded. Each scporate station file was then matched against a logically related file to obtain the number of v ehicles traveling from one station to another and t he interval of time required by each vehicle to accomp lish this movement Thus for example, che license places observed at Station 1 (1-275, sout h bound outside l ane ) were maoched against the plates observed at Station 5 (Ashley Stree t exit r amp). The difference between the insta!lt at which a given license was observed at t he upstream (I-275) stat io n and the instant at which that same plate was observed at the downstream s tati on (Ashley Street) is the trav e l time between t hose two stations for the vehicle c a rrying that license plate. The percentage of the license places passing a given camera location that can be successfully transferred from videotape images to a computer fale depends on the quality of the videotape recording and the method used to effect that transfer. If the videotape license pla te images are of very high qu.1lity, then processing by means of an automatic pla t e reader results i n a relatively high percentage of these images being t ransferre d directly to a computer fil e for subsequent ana l ysis. If, as was the case w ith many of the videotapes in this project, license plate images arc in poor focus or too dark (or bright) o r a re otherwise ill -suited for automatic reading, then other means o f "reading" these images must be employed. Wherever possible, the license plate images were automati cally "captured/ from the more comprehensive image of the vehicle and transferred to a separa te view from which they could be read by a human operator and entered into a computer fi)e. This process is referred to as "'semi-automatic review and is q u ite COSt effective when coupled with the system function known as t he "plate trigge r of the machine vision system. For those videotapes, or portions of tapes, for which the license plate images ''ere unsuited for automati c "capture," human operators read license plates captured directly from t h e original tapes. The semi-au tom:ttic r ev iew system i s set-up to very qui ckly review the captured license plate images and verify the visual i mage against the ASCII file data created by the machine visio n system This sec-up allows human operators to manually enter or change a particular machine visio n entry plate) from the image already stored in the database. This semi-automatic review system is very quick and, since the time and date stamp are automatically included with the licen.
PAGE 34

Demonstrati o n of Video-Based Technology for of T r affic Data Collection Field Demonstration Objective and Scope I n Fcbmary 1995, the Cente r for Urban Transpormtion Research (CUTR) joined w ith T rans fomation Systcms,lnc. (T ransfo) and Computer Recognition Systems, Inc (CRS) to demonstrate the use of video C\\rnera and machine vision technology for analyzing traff ic movements through t he interchange of 1 275 and 1-4 in Tampa, Florida. Licen s e plates were recorded at tbc two southbound lanes of 1 275, the two westbound lanes ofl-4, and at the Ash ley Stree t off ramp from 7:00am to 9:00am (or 7 : IS am to 9:15am) on three consecutive weekdays. Figure 3 illustrates the six camera positions uti l ized fort he field demonstration. The machi n e visio n system was then used to read these license p l ate images and, by matching identical images at different locations, to construct the pattern of origi n s and destina t ions through the interchange. Since the precise instant at which each license plate was recorded was known, it. was also possible to calcu l ate the travel times of vehicles from point to point The following photograph shows the license plate reading location (camera H5) at the Ash ley Street ex i t ramp A separate video camera was posit i oned (camera K6) to record the number of occupants in each vehicle exiting at Ash ley Street. 32 -===========Travel Time. Origin-Destination, and Average Vehicle Oct:upancy

PAGE 35

Demonstralion ofVideo-Based Technclogy for Automati<>n of Traffic Data Collection Figure 3: Camera Locations for Tampa Field Demonstration Project North CW) N.T.S. C- 1 4 Tampa Trawl Time, Origin-Destination, and A'VI!Tage Vebic/e Occupancy -=========== 33

PAGE 36

Demonstration o[Vitko-Based Technology for Aut o mation of Traffic Dala Collection The next photograph indica t es the camera perspective from position #6. T h ese video records were man ually analy7.ed and compared with obtained by human observ ers at this same lo cation. Camera #6 perspective at the Ashley Stt'tet exic ramp The overa ll obj e c tiv e o f this project was to p rovide a brief demo n stra tio n of an automated vi d e o-bas e d traffic d a ta collection system by monitoring t ravel patterns, travel time, and vehidc occu p ancy into downtown Tampa from t h e and 1 4 corr i dors The primary objective \ v as tO conduct a threeday long study of morning rush hourtraffic (peak two hours) from 1 -275 southbound and 1 4 westbou n d traffic entering downtown Tampa via Ashley S treet for origin destinat ion and travel times. The secondary obj ective was to cap t ure the vide o images ofthe vehide occupantS as they pass by the A shley Strectlocation an d to eva luate the feasibility o f using this o l ternativ e tochnology to measure the vehide occupancy of the traffic flow. O n the third da y, thel-4 c ameras (#3 and # 4 i n F igur e 3) were relo c ated approxim ately I I m i l es upStr eam to the Li vi ngston Avenue over pass o f l -275. T h e next two photographs show the Liv i ngston Avenue overpass, both along Living st o n A ''enue and looking down on l-275 southbound f rom the overpass. The secondary obj ective was to ca pture the vide o images of the vehicle occ u pantS as they pass by the Ashley Street l ocation and to eva l uate the feasibility of using this alternative techno log y t o measure the vehicle occupancy o f the t r affic flow In order to provide efficient control and fac ilitate t he coll ectio n of qua lity data, the pro ject scope divided i nto two m a jor p has es: (I) training and survey and (2) data ana lys i s. T his section describest he scope of wor k for each phase and the approach for comp l eting ea c h to achieve the over all proj ec t objective 34 ============= Travel Tim e, Origin-Destinati()n, and A veragt Vehicle Occupancy

PAGE 37

Demonstration ofVidei)-Based Technology for Automation of Traffic Data Collection Alternate camera locations #3 tmd If at LivitlgstonAvenue overpass Training and Surveys This phase consisted of performing the ground wor k needed to condu ct the actual au to mated t r a ffic data su.-veys and the overal l management of the project. This task consiSted of training the survey crews in the use and o p e ration of the video cameras used as well as the safety and the quality contro l measures to be adopted d u ring the performance o f the origin-desti nat io n surveys. C l assroom and field training sessions were cond u c ted train the six individuals perfonniog the v ideo-based s urveys. On-Site Pre-Survey Plarming Visit This task consisted of finali zing t h e following details: locatio ns for camera placement, exposure settings (direction of sun, shadows, etc.), which is continuo usly updated as needed with a "supervis ory camera," and l ogistics for camera operators. The onsite visit was performed by the technology vendor and CUTR personnel who were traine d and supervised by the techn ology v end or consultants to determine-the ideal locations for camera placement and logistics as well a s o t her pert in ent pan'tmeters (e g site a ccess personal safety requirements maintenance of traffic, and consistency of video-taping and camera clock sychr o nization) The photo graph on page 37 illust r ates camera readiness activities in the field required p rior to data collect ion Tra1;-e/ Ti me, and Average Vehicle Occupancy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 35

PAGE 38

36 Demonstration of Video-Based Technology for Automation of Traffic Dara Co/lecrion Camero perspectitlt looking down on 1275 routhbound outside Lme from Liv i ngston At:enut OiltrpdS$ -===========;;;; Travel Time, Origin Destimttion, and Average Vehicle Occ-upancy

PAGE 39

Demonstration of Video-Based Technology for Atttomalion of Traffu: Data Collection Pre-data collectwn ramera ch
PAGE 40

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Hi-8mm vidto camcorder umi in fzeld demtmurations 1 westbound, median-side lane (camera #4) Ashley Street exit, license plate view (camera #S) Ashl ey Stree t exit, windshield view for vehicle occupancy (camera #6) The surveys were performed during the morning peak two hours of traffic flow to indicate the morning rush hour patterns. Data Analysis Once the field survey was comp l eted, all t'Opes were brought back to CRS, where they were processed using the license plate reader The processing consisted of playing the tapes through a proprietary license plate reading system. The end results of this process was a database cons isting of license plate numbers and t h e ti me and location at which they were observed. This phase consisted of synthesizing the data collected in Phase I to develop the origin-destination mat rix. In addition to genera ting the origin-destination matrix, informa tion on travel times wer e develope d as a part of t h is phase. The origin destination matrix was developed through the use of a license p lat e matching algorithm. Apan from m:urix and travel time summaries, the data were used to e .valuate the feasibility of matching this da tabase to the database of video images to c h eck vehicle occupancy. 38 ;;;;;;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;:;;;;;; Travel Time, OriginDestination, and Average Vehicle OccupanC)'

PAGE 41

Demonstration Technology for Automation of Traffic Data Collection Discus sion of Results Comparison with Tub e Co11t1ts The v ol ume of traffic passing each ca mera station during the two hour s urveys was measured by t h e F lorida Depart ment ofT ransp<>rtation by means o f pneum ati c road tubes, as s hown in Tab l e I. A comb i natio n of m:anu3J and 1D.2.chine visi on proced ures were used to .. read,. t.he tapes, so the p ercentages of co.pture vary considerably from station to station depending up<>n the quolity of each videotape and the mix of techniques used to process a given t a pe. Ovetall, for the three days of videotaping (six houn), a wul of 36,233 vehicles passed by the six camera locations, and of these plates were reod {67.9 percent capture). Tab le 2 detoils the p l ate readi ng result s over the tbrccx!ay de m o nstrati on The tube cou n ts a t Statio n s 1 thro ugh 5 on February 2 2 and 23 wer e tak en as the base from which t o adjust the origind estination volu.ne, s. Trave l Time Results Meon travel times from Stations 1, 2, 3 an d 4 to Station 5 {a distance of 1.15 miles) are presented in Tob lcs :J-6 for February 22and Tables 6-9 for February 23. Given thot Stations 3 and 4 were relocated from 1-4 to Livingston Avenue (a distant<: of approximately 11 miles) on February 24, from-tO tabulations are shown in Tables 11-18 for movements from S t a t ions 1 and 2 to Stotion 5, an d from Stations 3 and 4 to Stations I a n d 2 Mean travel times, st an d ard deviat ions, a nd coefficients of vari at i o n are presented for IS minut e time interval s on February 22 an d for 30-minute time intervals o n the 23 and 24. In most in s tances b oth the mean travel times and the coeffic ientS of vori ation appear to be q uite reasonable Tuvel t imes vary generally with the ebb and flo w of tnffic volumes over the 7 :15am to 9:15am period of observation A good exampl e of this is seen i n Table 3, where travel t im e averages 96.4 seconds over the7:15 am to 7:30am interval, risesqui cldy t o 121.0seconds by 8:00am, remains high uotil8:45 am, and then drops to 87.3 seconds by 9:15a .m. In Tab les 4 5, and 8 the highlighted sect i ons identif y time periods for w hich time in dexes on t h e field ta pes were, mo s t l ikely, impr operly recor ded foll owing a n imcrnapt.io n in ta ping. T h e coefficient- of varia tion associated with the vari o u s travel t ime estima tes g enerally s u p p o rt the co n clusion that these estimates accurately r eflect t h e true vnlue of mean travel time f or eoch time per i od. A total o f 5 9 o f the 76 coefficients of variatio n are below 15 percent, and o nl y 9 are a bove 25 percent. Given the variability inherent in travel spoeds in both freely flowing and interminendy congested traffic, these sm.Jl variations around estimated mean travel time. is reassuring. Trawl T ime, Origin-DeJtimttlon, and At>erage Vehicl e Ow
PAGE 42

40 Demonstration of Video-Based TechnoU>gy for Automation o[ Traffic Data Collection Table 1: Traffic C o unts (Vo lum e) for Two -H our Peak Period During Video-Based Survey Camera Location Dare Count 3 Day Avg. CV(%) I l-275slb-out 02/22195 2,728 2 l-275slb-in 02/22195 2,395 3 l 4w/b-out 02122/ 95 3,613 4 l -4wib-in 02n2i95 3, 1 02 5 Ashley St. 02n2!95 2,502 I 02123195 2,776 2 lslb in 02 / 23/95 2,465 3 l 4wlb out 02/23 i 95 3,448 4 14 wlbin 02123i9S 2 ,881 5 Ashley St. 02n3/ 95 2,417 I l 275stb-out 02n4!95 2.431 2 l-275slb-in 02 / 24/95 2,089 3 02124/95 726 4 1 75slb -in 02/241'95 278 5 Ashley St. 02/24195 2,382 -. . . . CV c o efficte n t of vanatlon (ratio of standard dev1atton to the mean) sfbin ... sourh bound inside lane s.rb-out = south bound outside lane wll>-in =west bound inside lane wlb-out = west bound outside Jane 2.645 2 ,316 3, 391 3)332 2,434 2 645 2,316 3 ,391 3,332 2, 434 2,645 2 316 771 269 2 ,434 Table 2: License Plate Reading Results Date Tot al Plates Total Plates Rtad 02122/95 1 4 ,3 4 0 9,070 02/23/95 13,987 8,875 02124195 7,906 6.681 TOTALS 36,233 24,626 5.8 7.0 6 1 14.7 2.1 5.8 7 0 6.1 14. 7 2.1 5.8 7.0 4.2 5.0 2.1 Percent Read 63.2% 63.4% 84.5% 67.9% -============ Tra7;el Time, Origin-Destination, and A'Lerage Vehicle Oc,upancy

PAGE 43

Demonstration of Video-Based Technology for Atttomation of Traffic Data Collection Table 3: Mean Travel Time and Standard Deviation by IS-Minute Intervals (from l-275 SB outside lane to Ashley Street exit ramp, February 22, 1995) -' 'time #of Matched Mean Tra\el Standard Coefficient of (AM) l'lates Ti m e (see) Deiation (see) Variation (%) 7:15-7:30 1 5 96.4 10.8 11.2 7:30 7 :45 42 98.4 11.4 11.6 7:45 -8:00 28 1 21.0 16.6 13.7 8 :00-8:15 1 4 120.1 10.5 9.0 8:15 -8:30 1 6 111.9 1 2.5 1.1.2 8:30-8:45 1 9 123.8 16. 9 13.7 8:45-9:00 20 87.5 7.7 8.8 9:00-9:15 1 2 87. 3 7.2 7.8 Table4: Mean Travel Time and Standard De\'ia tion by 15-Minute Intervals (from l -275 SB inside lane to Ash ley Street exit ramp, February 22, 1995) Time #of Matched l'vlean Travel Standard Coefficient of (AM) Plates 'fime (sec) Deviation (sec) Variation(%) 7:15-7:30 6 100. 8 7 9 7 8 7:30 7:45 6 109.5 4 1 3.7 7 : 4 5-8:00 3 129. 7 8.1 6 7 8:00-8:15 4 119. 0 11.9 1.0.8 8:158:30 7 110.3 11.6 10.5 8:30-8:45 9 79.3 20.1 25.3 .. -. 8:45-9 : 00 6 28 5 6io . .."21 \ -' 9 : 00-9:15 8 25.6 9.9 -.. ,., -. . Note: shaded area designaJes problem wilh time stamp. Travel TiTTU!, Origin-Destination, and Average Vehicle Occupancy =========== 41

PAGE 44

Demonstration of Video-Based Technology for Automati o n of Traffic Data Collection Table 5: Mean Trave l Time and Standar d De v iati o n by IS-Minute l nten-als (from 1-4 WB outside lane to Ashley Street ex i t ramp, February 22, 1995) Time #of Matched Mean Travel Standard Coefficient o f (AM) Plates Time (sec) Deviation (sec) Variation (0/o) 7:15-7:30 1 5 90. 7 5.8 6.4 7:307:45 28 92.1 11.9 12.9 7:45 8 : 00 II 93. 2 9.1 9 8 8:00-8:15 8 75. 1 8.3 II. I 8:15-8:30 13 72. 2 12.1 16.8 8 : 30-8:45 9 68.9 13.2 19. 2 8 : 45-9:00 II 43. 6 1 3.5 31.0 9:00-9:15 6 3 2 .5 12.3 37.8 N()le: s haded area designates problem with time SIOmp. Table 6: Me a n Travel Time and Sta ndard De viation by 15-Minute Intenals (from 1-4 WB inside lane t o Ashley Street exit ramp, February 22, 1995) Time #of Matched Mean Tra\'el Standard of (AM) Plates Time (sec) Deviation (sec) Variation (%) 7: 15-7:30 18 82. 8 32.1 38. 9 7:30 7:45 1 9 107.4 15.9 14.8 7:45-8 : 00 9 147.6 32.5 22 .0 8:00-8: 15 8 1 60.0 7 7 4.8 8 : 15-8:30 12 154.5 8.6 5 8 8 : 30-8:45 1 2 149.3 16.4 11. 0 8:45 9:00 10 120.9 19.4 16.0 9:00-9: 15 II 102.3 14.6 14.3 42 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Travel Time Origin Destindti
PAGE 45

Demo>ISlration of Video-Based Tech11ology for Automatio>l of Traffic Data Collection Table7: Mean Travel Time and Standard Deviation by 30,Minute Intervals (from 1 -275 SB outside lane to Ashley Street exit ramp, February 23, 1995) Time # ofMatthed ?\'lean Tnwcl Standard Coefficient of (AM) Plates Time (sec) Deviation (sec) Variation(%) 7:15-7:45 58 120.1 17.0 14.2 7:45 -8; 15 7 9 150.4 10.4 6.8 8:15-8:45 1 2 7 154.9 11.2 7.2 8:459:15 99 1 55.7 6.4 5.2 Table 8: Mean Travel Time and Standard D eviation by 30-Minute Intervals (from 1-275 SB inside lane to Ashley Street exit ramp, February 23, 1995) Time #of Matched :Mean Tra\el Standard Coefficient of (AM) Plat .. Time (sec) Dniation (sec) Variation (%) 7:15-7:45 13 103.2 10.6 10.3 7:45 8:15 26 1 09.4 14. 2 13.0 8:15-8:45 41 86.2 19.1 22.2 i: r ::;id"!.'( j .... Note: shaded area nul. "" J Table 9: Mean Travel Time and Standard De,iation by 30-Minute Intervals (from 1 -4 WB outside lane to Ashley Street exit ramp, February 23, 1995) Time #of Matched Mean Travel Standard Coefficient of (AM) Plates Time (sec) Deviation (sec) Variat ion (%) 7:15 -7:45 24 106.0 16.0 1 5 1 7:45-8:15 29 118.3 26.4 24.0 8:15:45 34 138. I 7.7 5.6 8:45-9:15 54 141.9 6 9 4.9 Travel Time, Origin-Destination, and Average Vebick Occupancy =========== 43

PAGE 46

Demonstration of Video-Based Technology for Automation of Traffic Data Collecti071 Table 10: Mean T ravel Time and Standard Deviation by 30-Minute Intervals (from 1-4 WB inside lane to Ashley Street exit ramp, February 23, 1995) Time #of Matc hed Mean Travel Standard Coefficient of (AM) Plates Time (sec) Dev iation (sec) Variation(%) 7 : 15 7:45 25 103.0 1 3.3 12.9 7:45-8:15 2 1 115.2 9.1 7.9 8:15-8 : 45 30 134.0 12.5 9.3 8:45-9 :15 21 137.4 9.7 7.1 Table 11: Mean Travel Time and Standard Deviation by 30 Minute Intervals (from 1-275 SB outside lane to Ashley Street exit ramp, February 24 1995) Time #of Matched MeanTra>el Standard Coefficient of ( AM) P la tes Time (sec) DeYiation (stc) Variation(%) 7:15:45 106 106.5 11.2 10.5 7:45 8:15 145 115.2 43.3 37.6 8:15-8:45 118 1 37 0 9.3 6.8 8 :45 -9: 15 90 232.5 1 29.2 55.6 Table 12: Mean Travel Time and Standard Deviation by 30 -Minute Intervals (from 1 -275 SB inside lane to Ashley Street exit ramp, February 24, 1995) Time #ofMatchcd Mean Trnel Standard Coefficient of (AM) Plates Time (sec) Deialion (sec ) Variation(%) 7:15-7:45 57 122.4 15 4 12 6 7:45-8:15 56 140,4 11.3 8.0 8:15-8 :45 84 I 51.8 1.9 5.2 8:45-9:15 55 256. 2 143.0 55.8 44 ============ Travel Time, Origln-Destination. and Average Vehicle Occupancy

PAGE 47

De11W115tTati
PAGE 48

Demonstration of VidM-Based Technology for Automation of Traffic Data Collection Table 16 : MeJtn Travel Time and Standard D eviat i on by 30-Minute Interval s (fro m 1 275 SB in side lane @ Livingston Avenue ove rp ass to I-275 SB out s id e lane @ l 4 / 1-275 Interchange, February 24,1995) Time #of Matched Mean Travel Standard Coefnclent of (AM) l late.t Time (sec) Deviation ( s: ) Variation 1 :15-1:45 39 1,151.7 104.6 9.1 7 : 45 :15 23 810.4 155.6 19. 2 8 : I 5 -8 : 45 1 8 688 6 30.6 4 4 8 : 45-9 : I 5 16 574.5 27.8 4 8 Table 17: Mean T rave l Tim e and Standard De via tion by 30-Minute Intervals (from l 275 S B @ L ivingston A venue overpass to I -275 S B @ 1 4/I -27 5 Interc h ange, February 24, 1995 ) Time #of Mat c hed Mean Tra vel Standard Coefneient or (AM ) Plates Tim(>) Deviation (s .. ) Variation (%) 7 :15-1:45 1 0 8 1,194.1 114.9 9.6 7:45-8:15 86 944.7 174.9 18.5 8:15-8:45 68 796.2 111.6 14.0 8 :459:15 48 672. 2 131.7 1 9.6 Tabl e 18: Mean T ravel Time and Standard Deviatio n b y IS-Minut e Interv als (from 1 -275 S B @ Li vingston A ve n ue overpass to 12 7 5 SB @ 1-4/I-275 Inte r change February 24 1 995) Time #of M a t ehed M ean Tra\l'tl Standard CO<'Mcient of (AM) Plates Tim e (sec) Oevhttio n (sec) Variation(%) 1: l 5-7:30 60 1 ,143. 4 108.7 9 5 7:30-7:45 48 1.257.5 88.3 7 0 7:45-8 : 00 3 8 1,012. 3 129.1 12.8 8 : 00.8: I 5 48 891. 1 188.6 21.2 8 : I S-8:30 36 826. 7 136.1 16.5 8 :308:45 32 761.9 60.9 8.0 8 : 45-9:00 37 690. 1 143.0 20.7 9:00-9:15 l l 6 11.9 51.1 8.4 46 =========== Tr
PAGE 49

Demonstrario11 Tecbnology for Automatio n of Traffic Data Coll tction Origi n Destination The co mputation of the distributi o n of traffic move ments b etwee n t h e various camera stations required that each 0-0 pair volum e be adj u sted according to the product of the percent o f vehicle passing each .. socia ted stat ion pair and thot the sum of the movemen ts thus comp uted be further adjusted tO comply with t h e tube count volume at the destination station. This prOCC$S is illustrated in the f ollowing example for the movements from Stations 1, 2, 3 and 4 to Station 5 on February 22. 1-275 SB outside lane to Ashley Street (1-5) !62 X 0 39 X 0.41 1013x 2,502 tube COIInl dt #5 2SQ2 1840 where, 162 = tbenumberofmAtchedplaus 0.39 = percent of pl4tet ntad at #1 0.41 = percent of p/.tts ntad IIJ 1,840 -total number of pia us at # 5 1-275 S B inside lane to Ashley Street (2) 1-4 WB outside lane to Ashley Street (3) l-4 WB inside lane to Ashley Street (4-5) 48 X !01 X 99 X Totals 0.56 X 0.41 1 0.79 X 0.4 1 0.79 X 0.41 209x 312 X Jll!l X 1840 X 2502 1840 22QZ. 1840 1840 1.359 1378 (55%) 284 (11%) 424 (17%) ill (17%) 2502 (!00%) Calcu l ations for De stina tions of Entering Traffic at Camera Stations 1 Through 4 1-5 U78 + 2728 = 50% of traffic from I to 5 2-5 -284 + 2395 = 12% of traffic from 2 to 5 3-5 424 + 3613 = 12% o f traffic from 3 to 5 4 5 -416 + 3445 12% of traffic from 4 to 5 The distribu tion o f ori&ins to deStina tio ns for Fe bruary 22 and Feb ruary 23 are summ arized in T ablcs 19 22. No ori&in-
PAGE 50

Demonstration of Video-Based Technology for Auromation of Traffic Dara Collection V thic/e Dar a A video camera # 6 ) was setup s eparately from the previous five l icense plate video comems to enab le the viewing of pas senger> through the front windshi e ld o f each passing vehicl e. There was no odditio nal source lighting used during this survey, but only the aV2ilable y February 22, 1995, are shown in Table 23. As shown three fourth s of the vehicles are passenger car>, and almost 85 percent of the vehicles have o nly one passenger. The averag ooocupa ncy rate was 1.17 during the Wednesday morning rush hour. However, from various minut eby-minme reviews o f the videotape, it was shown that the average occupancy rate ranged between 1.13 and 1.24 peop l e per vehicle (illustrated later in thisreport). T he video system was able t o pick up, oo a real time basjs, the actual fluctuations in vehicle occupancy. Table 19: Origin-Destination Matching Study and Through Traffic Analysis (February 22, 1995 from 7:15am to 9:15am) Came"' I of Vehi cles I or Vehldes Actual I of Nonnalized Vebldes Through Pair a t First at Last Matclltd Number or Turning Trame (1-275 SB) Camera Camera License Mattbtd (%) (%) (1-4 WB) Location Location Plates Vehides 1 to 5 2,728 2,502 162 1,378 50 (1 275 SB) 73 2 to 5 2 395 2,502 48 284 12 3 to S 3 ,613 2,502 101 424 12 (1-4 WB) 88 4 to 5 3.102 2,502 99 416 12 Table 20: OriginDestination Matching Study and Through Traffic A nalysis (February 23,1995 from 7:15am to 9:15am) C amera #of Vehicles fl of Vehicles Actual# of N ormalltd Vehicles 11ll'ougb Pair at First at Match ed Numbtrof Turning Trame (1275 SB) Camera Camera License Matched (%) (%) (1-4 WB) Location Location Plates Vehid,. I to 5 2,n6 2,417 162 1.366 49 (1-275 SB) 67 2 to S 2,465 2,417 48 370 15 3 tO 5 3.448 2,417 101 396 II ( 1 -4 WB) 89 4 to 5 2,881 2,417 99 285 10 48 ======= ===== Travel Time, Orig;nDnlinatioll, and Atterage Vtbiclt Occupancy

PAGE 51

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Tab le 21: Origin of Traffic Exi t ing Ashley Str eet (Camera Station 5) (February 22, 1995 from 7:15am to 9: 15am) Camera Pair Number of Number of Actual Normalized Exiting (1-275 SB) Vehicles Vehicles Number of Number of Vehicles ( 1-4 WB) at First Camera at Last Camera Matcbed Exiting (%) Location l-ocatio n License Plates Ve-hicles I 10 5 2 728 2.502 162 1,378 55 U-27SSB) 2l05 2,395 2,502 4 8 284 11 31<)5 3,613 2,502 101 424 17 (1-4 WB) 4 10 5 3 ,102 2,502 99 416 17 Total Vehic les Exiting at Station 5 2,502 1 00 Table 22 : Origin o f Traffic E x iting As h ley Street (Camer a Station 5) (February 23, 1995 from 7:15am to 9:15am) Camera Pair Number of Number of Actual Normalized Exiting (1 SB) Vehicles Vehicles Nu.mber of Number of Vehicles (14 WB) at First Ca m era at Last Camera Matched Exiting (%) Loeation Location License Plates V e hicles J to 5 2,776 2,4i7 162 1,366 57 (1 NB) 2 to 5 2,465 2,417 48 370 15 3 t o 5 3,448 2 ,417 101 396 16 (14 \VII) 4 to 5 2,881 2,417 99 285 12 T otal Vehic.Jes Exiting at StatJon 5 2,417 100 T a bl e 23: Occupancy of Vehicu lar Traf fic Exi t ing Ashley S treet (Camera Stat ion 6) Trucks Tractor Buses Total Cars &Vans Trailers &Other Vehicles 1,754 545 27 2 2,328 75.3% 23.4% !.2% 0 .1% 100% Vehicles Vehicles Vehicles Vehicles Tota l with l with 2 with 3 with 4 + Occupants 1,994 333 26 8 2,720 Anrage OccuPanc)' Rate: 1.17 Travel Time, and Average Vebicle ===========-49

PAGE 52

Demonstration of Video-Based Technology for Automation of Traffic Data Collection This vast amount and specific type s of dat:.t were p reviously unavailable to t raffic managers, who rely on manual counts of vehicles and occupancy Additi o nally, since this syst e m captures l i cense plate numbers automatica lly, i t gives traffic manage.rs the ability to identify and contact (survey) drivers who are car pooling. This capabil ity is helpful in establishing better methods for e ncouraging pe ople to double and triple up i n theirs cars, l i ght trucks, and vans Site Visit Day O n T h ursday, February 23, 1995, CUTR representativ e s and t he technology vendor team conducted a site visit day to a llow interested parties to observe t he video-based traffic data collection activities of the am rush period Several area consultants and public sector trans portation officials attended t his field demonstra tion. T h e f ollowing photograph depicts visitors who attended the site visit day from Hillsborough and Manatee Counties representatives. Following the morning data coJiection, visitors were also invited back to CUTR's office for further demonstrations of the video-based technology, and to answer any questions. The purpose of the site visit day was to expose interested transportatio n professionals to this type of ap plication first -han d, and enc ourag e further appl icat ion of this technology to traffic performance data collection. monitoring. -and analysts. Sitt visit 50 -===========Travel Time, On'ginDtstination, and Average Vehicle Occupancy

PAGE 53

Demonslrarion ofVidtoBIISed Technology for AHUJmation of Traffic IJdta Collection Video-Based Survey vs. Visuol Observation Findings Avcrngc travel times were obtained and used to colculatc avernge speed of trnffic flow between camera l
PAGE 54

Demonstration of Video-Bas ed Technolo gy f o r Automation of Traffic Data Collection ... !!l c .. :.;: .. e . '" .. > f ... Figure 5: Comparison of Traffic Data Collection Results-Average Travel Time (from 1 -275 outside lane to Ashley Street Exit Ramp) 10 9 8 7 6 s 2 0 1.61 I.SII 7 : IS 9 .35 7 :30 7:45 1 OVideo -Based Visual Observation Note: Average Travel Times from Visual Obser.ations are based on t S total license plate matche S for all time periods.. Average travel 1imes from Video .. aased Observations are based on total16S license plate matches lAS 8 :00 8 : 008 : I S 8:15-8:3 0 8:3 0 -8 :45 8 : 459:00 9 :00-9 :15 Time of Day (Wednesday, Februay 22, 1995) 56 percent of traffic originates from l 275 southbound (outside lane) 13 percent of traffic originates from l 275 southbound (median-side lane) 17 percent of traffic origina t es from 1 4 westbound (outside lane) 1 4 percent of traffic originates from l-4 westbound (median side lane) Figure 6 illustrates t he origin d ist ribution patterns observed on day 1 (Wednesday, Fe b ruary 22) and day 2 (Thursday F ebruary 23). Figu r e 7 h i ghlights t he number of matched license plates, used for origin destination determination, for video-based versus visual observations comp a red to the normal i zed origin destination volumes factored from the machine counts. I n all instances, vi deo -ba sed mat ches far exceeded the matches obtained by visual observation. As noted previously, the average ve hicl e occupancy (AVO) rat e was manually obtaine d by review of t he Only one day (Wednesday, February 22) was compiled although all three days were videotaped T he overall average vehicl e o c cupancy for this day was 1.17, although the findings are even more beneficial if broken down into IS-minute intervals (which the video tape record allows). Figure 8 illustrates t he var iation of AVO on February 22 52 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Travel Time, Origin-Destination and Average Vehicle Occupancy

PAGE 55

D1!111onstrati
PAGE 56

"' u = .. "' !.: 0 .. "" .; !! .. < Demonstration of Vitko-Based Technology for Automation of Traffic Da14 Collection 1 .24 1 .11 1.1 1.18 1.16 1.14 1.12 I. I 1.08 !.06 Figure 8: Average Vehicle Occupancy Data 1.24 7 :15-7:30 7 :30-7:45 7:45-8:00 8 :00-8:15 8: 15-8:30 8:3().-8:45 8:45-9:00 9 :00-9:15 Time or Day (Wednesday, February 22, 1995) Qu ite surprlsingly. average vehicle occupancy measurements showe d little variation comparing videobased versus visual observation (even though video-based had 74 p ercent greater capture rate of all passing vehicles). Figure 9 portrays th i s s i milarity between the two methods of AVO data collection. At this time, it is unknown whether thls similarity may be purely coincidental or The final comparison between video-based and visual data collection methods for this field demo n stration is illustrated in Figur e 10and 11. These figurescomparethecollection and processing time per usable survey and the relative cost per usable survey, respectively. As can be seen from thes e figures, collection and processing time is about eight times greater for the visual technique, and onl y about 50 cents less per usable survey. Using Video Imagery to Measure the Quality of Traffic F low The large number of detailed observations of ind ividual vehicle movements recorded on video tape provides not only accurate measures of travel times and origin-destination exchanges, but also intriguing insights into the quality of traffic flow on the highway segment under study. Traffic engineers generally characterize the quali ty of traffic flow along a given highway segment in t erms of "level of service," the essence of which is the degree of freedom of movement enjoyed by motorists in that flow. When traffic 54 ;;;;;;;;;;;;==========-Travel Time, OriginDestination, and Average Vehicle Occupancy

PAGE 57

l>em1S S:OO a.-ooS:IS D Visual Observation 1.23 I : IS 1:3 0 1.23 1 :30-S :IS 9:00 Time or Day (Wednesday, February ll,1995) --9:1S volume is very light, each motorist is free to drive at a speed that is essenti ally unaffected by the presence of other vehicles. As traffic volumes increase, the speed at which each driver operates is increasing l y iofiuenoed by the presence of other vehicles. The average speed of traffic may not diminish appreciably, but the opportunity for any given driver to operate at speeds higher or lower than the average is prog=sively diminished Conventional traffic measurement techniques typicall y collect data from whi ch only average traffic flow characteristics can be derived: average speeds and travel times; flow rates; a\erage density ; and other related measures of flow qualit y. In contraSt, videotap ed records of traffic p rovide detailed, vehicle by ve hicle, observations from which both average flow characteristics and fine-grain variations with these average parameters can be determined. Previously summarized results in Table 3-18, in which both mean IS minute or 30 minute mvcl tim .. are shown along with the coe fficient of variation associated with each mean travel time, exemplify this unique ability of vidco12ped traffic observations. The coeffic ient of variation is a statistical measure of the variability of the individual vehicle travel times that together make up the average travel time for a given time interval. This measure of variability is under certain circumstances, associoted with the freed om individual drivers have to select their own opera ting spee d At very low volumes, the range o f individ ual choice is high and t h e coefficient of vari2tion should be relatively high as well. Trawl Time, Origin-Destination, and Aflt!Tage Vehicle OccNpttncy ============55

PAGE 58

56 II -" = i s .. I: .!! 0 Q = Demonstration of Video-Based Technology for Automation of Traffic Data CoUection Figure 10: Collection and Processing Time per Usable Survey 8 7 8 1 fNote: Video-Based Observations are based on 1,125 matched license plates and 2,328 6 vehic l e occupan<:ies. Visual obse{'\'at ions are based on 124 matched license plates and 583 s ve-hicle occupancies. o Vi4eo--. 0;$;" 4 Visual Obscnatio n 3 2 0 9 ou_ __________ _L __________ 4 4 4.:14 4.3 4 l 4.1 4 3.9 3 3 3.7 3 6 3 5 Type of Observation Figure II: Relotive Cost Per Usoble Survey Note: Video-Based Observations are based on 1,125 matched license plates Md 2,328 vehicle occupandes. Visual observations arc based o n 124 matched license plates and 583 \'eh ltle OVideo-Bascd Visual Observation 3 8 1 Type of Observatio n ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Travtl Timt, Origin-DestintltWn, and Average Vehicle Occupancy

PAGE 59

Demonstration of Video-Based Technowgy for Automation of Traffic Data Collection At low mean travel speeds should b e h i gh also. As traffic v olumes increase, the range of choice of individ u al drivers is r educed and, as l ong as traffic" con tinues to move smooth ly--b u t a t a s l ower pace the coefficient of variation shou l d also decrease, signaling a reduction in level of service. F i nally as volume and, more importantly, densit y point wher e traffic flo w abruptly changes from the laminar t o the turbulent engine regime, strea.ffi itsdf becomes highly variab l e and traffic moves in fits and starts at a low average speed. At this pomt, flthough t h e freedom of individua l drivers to choose their own operating speed is essentially reduce d to zero, the increased variabili ty i n the stream flow results in a concomitant increase in the co efficient of variation, which, in the range o f extreme congestion, signal s yet further red uction in level of service. The data that were collected in this demonstration project were not intended to explore the relations hip between level of service and voriation i n travel t i m e described above. Although there i s some evidence in Table J-18 t hat the wefficien t of variation is h ighest during those i nte rval s when mean travel t imes are highest t h e number o f observations on which such as association is based is coo li mited to provide any statistical validation Neverthe less, the potential o f video traffic flow analysis to reveal the details necessary to capture fully the quali t y of flow and, thu s to provide in f ormation essential for congest i o n management and assessment is obvious As f ur the r example o f the type of detailed vehicular movement information that can be developed from videotaped records, con s ider Figures 12 and 13. These figures are based on data collecte d on Route 520 i n Seattle.21 They show the travel times for individual vehicles arraye d in order of their arrival at the upstream camcorder observation si te. The int eraction of one vehicle wit h ano th er is evid e n t in the waveform of th e graph, wh i ch clearly illu strates the p l atooning effects to v1hich i ntervehicle influences would lead. Wit h data such as this, i t would b e possible to develop congestion monitoring and m a nagement procedures based on indicators other than gross averages of traffic flow quality. New Traffic Performance Measures From the data collec ted dur ing t h i s field demonstration, more meaningful traffic performance measures can be derived. Thr e e such measures of e ffectiveness are offered for consideration. For example, Figure 14 illustr ate s 15-minute interva l s for volume vs. average vehicle occupancy at the Ashley Street exit ramp for Wednesday,February 22. Similarly, Figure 1 5 depicts 15-minut e interva l s for volume vs. average travel speed for the I-275 southbound outside lane to the Ash ley Street exit ramp (1.1 miles t h rough the interchange). Finally, based on average vehicle occupancy rates and volume, total person-tr i ps can be det ermined as shown i n Figure 16 (Ashley Street ex. it ramp f or Wednesday, February 22). Travel Time, Origin-Deltinttti()>t, <17td Average Vehicle Occupancy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 57

PAGE 60

58 Demonstrari;m of Video-Based Technology for Automation of Traffic Data Collection i 1! i l l 1 i l Figure 12: Array of Travel Times for Vehicles Entering and Exiting Lane I (ordered by time of arrival at upstream observation site) 02;01 01:44 01:2$ 01:09 00:52 00:35 00:17 .. ooL--------------------' 0:00 :00 O:OS:-'8 0:08:38 0:11:31 0:14:24 Figure 13: Array of Travel Times for Vehicles Entering and Exiting Lane 2 (ordered by time of arrival at upstream observation site) 02:01 0 1 :4.4 ou 01: 09 ... OO:M OO;t7 00 00 00:00 oft ., t .... -...... .. t '' r . .... . ., .. ., . . U :31 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;::;: Travel Time, and Awrage Vehicle Occupancy

PAGE 61

Demorntration of Video-Based Technology for Automation of Traffic Data Collection J S O 300 2SO J 200 ISO 100 so 0 350 300 250 .. 200 a .e 0 ISO > 100 so 0 Figure 14: Volume vs. AVO I =volume -.-AVO I l;l"'= r-' IT -.._ v 7 :157:30 7:30 7 :4S 7:4 S 8 : 00 "" "' 8:00. 8 : 1S-8:1S Se30 Time of Day 8:308 :45 8:45-9:00 F igure 1 5: Volume vs. Average Trave l Speed 7:. 1 S -7 : 30 ....._ 7 :30 7 :45 I'=volume ....-Average Travel Speed !-... 1:45 8:00 ..... 8:00-8:1S8 : 1 5 8:30 Time of Day -/ 8 :308 : 45 8:45-9:0!1 ,,t 9:00-9ol S 9 :00-9:15 1 .24 1.22 1.2 ).)8 1.16 1.14 1.12 1. 1 1.08 1.06 f so 4S 40 35 30 25 20 IS 1 0 s 0 0 ., z !J; : $. :. .. ... .. < Travel Time, Origin -DI!Stination, and Av.rage Vehicle Occupancy -==========59

PAGE 62

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Figure 16: Total Person-Trips 450 3 9 3 401 400 350 300 ... g 250 ... .. 200 6 z ISO 100 so 0 1:15 7:30 7:45 8:00-8:1S 8 : 3 0 8:45 9 : 00 7 :JO 7 : 4 5 8:00 8:15 8 :30 8:45 9 :00 9 : I S Time or Day 60 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Travel Time, Origin-Destination, and Average Vehicle CXcupancy

PAGE 63

Demonstrdtion ofVide()-Based Tecbnology for A11tomation of Traffic Data Collection Conclusions Given th e findings of t his field demonstrat ion project, automation o f traffic data gather ing an d a n a lysi s is feasibl e t h rough video and machine visi on technology application. This t ype of ITS technology satisfies the continuous m on itoring need oi congestion management systems. Meani ngf u l traffic data can b e collected and a n alyzed more often in a time l y and costeffective m a nner. Further as a tool for ClvlS performance monitoring, the traffic data collected in t h i s investigation (i.e., trave l time, origindestination, and average vehicle o ccup a ncy) can al s o be u tiliz e d to dev e lo p m ore m e a n ingful traffic performanc e measures or measures of effectiYeness (MOEs ). Travel Time Origin-Destination, and Average Vehicle Occupancy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 61

PAGE 64

Denumstrati
PAGE 65

2 Demonstration of Video-Based Technology for A1ttomation ofTraffic Data Collection Endnotes T. Cre=y and A. Dominguez, "De velopment of a Prototype Integrated Management System, ITE International Conf erence on Technology Tools for Transpomtio n Professionals. Moving Into the 21st Century, Ft. Lauderdale, Florida, April 1994. JHK & Associates, In c., "Hillsborough County Co ngestion Management System Work Plan, Orlando, Florida, September 1994. M.C. Pietrzyk "An Applic atio n of ns T cchnology for Co ngeion Management Systems," Proceed ings of ITE 65th Annual Meeting, 1995. W. Youngblood, "Integratio n ofT raffic Opera tions and Traffic Dat a Collection Nationa l Traffic Data Acquisition Confere nce, Rocky Hill, Connecticut, Septe mber 1994, p. 344. Institute ofT raosportation Engineers, "Manuol ofT raospomtion Engineering Studies, Prentice Hall, Cliffs, NJ, 1994, p.34. S. Turner, et al., "Con gestion Management: Data Collection Methodologies, Texas Transportation Institute, 1994. 1 10 Institute of Transportation Engineers, "Manual o f Trans portation Engineering Studies, Prentice Hall, Eos)ewood Cliffs, NJ, 1994, p. 111. Ibid D. Muntean, Jr., "OriginDestinati on Travel Surveys in the 1990'sUs ing Microcompu ters, Institu te of Transportation Engineers, Florida Section, 199;). T. Zakaria, "Cordo n Lin e Highwa y Survey for the Delaware Valley Region, Tran sportation Research Record, #1>05, Trans portation Research Board, Washington, D.C., !991, p. 97. II Ibid 11 D. McKinstry and L. Nungesser, An Evaluation of On-site Administered Origin-Destination Survey Methodologies: Postcar d Mail-Back vs. Interview," Proceedings of the >rd Nation,.! Conference on Transportation Planni ng Methods Applications, April 1991. ' M Schaefer, "License Plate Matching Surveys: Practico.llssues and Statisticol Cons i deration, Institute of TrAnsportation Enginee-rs journal, Vol. 58, N o.7, Ju ly 1988, pp >7-42. Transfomation Systems, Inc., P roposal tO Cen te r for Urban Transpo r totion Research for Field Demonstrotion Projec t: Automated Video-Based Traffic Data Collection," Houst on, Texos, january 1995. 63

PAGE 66

Demonstration of Video-Based Technology for Automation of Traffic Data Collection K. Miller, et a l. "Using Video Technology to Conduct 1991 Boston Region External Cordon Survey," T ransportat ion Research Record #1412 T ransportation Research Board, Washington, D .C., 1 993, pp. 46 56. T. Liu "Fi eld Tests of Trave l Time Survey Methodologies and Development o f a Stand ardized Data Processing and Reporting System," Natio nal T rnffic Data Acquisition Conference, Rocky Hill, Connecticut, September 1 994, p. 401. 1 1 Institute ofT ransportation Engineers, "Manual of T ransportation Engineering Studies," Prentice Hal l, Engkvood Cliffs, NJ, 1994, p 121. K Manges, et al., "Comparison of Home T ravel Survey Collection Methodologies, Transportation P la nni n g Methods Applications Third National Conference, Dallas, Texas, 1991. E. Peterson and J. Hamburg, "Travel Surve ys : Current Options," Transportation Research Record #1097 Transportation Research Board, Washington, D.C., 1 986, p.2. R. Gaulin, "A Procedure to Calculate Vehicle Occupa ncy Ra t es Using Traffic Accident Data: N ational Traffic Data Acquisition Conference, Rocky Hill, Connecticut, September 1 994, p. 354 Paul W. S h u ld i nc r eta!., "Detennining Detailed Origi nDestinat ion and Trave l Time Pattern Using Video and Machine Vision License P lat e Matching," presented at the Transportation Research Board Annual Meeting, Washington, D .C. January 1996 64 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;= Travel Time, and AverAge Vehicle Occupancy

PAGE 67

Demonstration of Video-Based Technology for Automation of Traffic Data Collection References Anderson, Craig A., e t al. "Cost-Benefit Analys i s of Video-Based Vehicle Detection Proceed ings of 6th International Vehicle & Information Syst e m s Conference, Seattle, Washington, J ul y 30-August 2 1995. Chatzi i oanou, Alypios, et al. "Video Image Processing Systems: Applications in Transportat i on." Proceedings of the 6th Internationa l Vehicle Navigation & Inf ormatio n Systems Conference, Seattle, Washington, Ju l y 30-Augu st 2, 1995. Creasey, T and Dominguez A. "Development of a Pro to type Integrated Management System.'' ITE International Conference on Technology Tools for Transportation Professionals Moving Into the 21st Cen t ury, F t Lauderdale, Florida, April1994. D' Agostino, Salvatore A., and Shuldiner, Paul. "App licati on of Vid eo/Mac hin e Vision Technology in Traffic Data Ana ly s i s." United S tates Department o f T ransport a t ion, Volpe Transportation Cent e r, 1994. Gaulin, R. "A Procedure to Calculate Vehicle Occupa ncy Rates Using Traffic Accident. Data." National Traffic Data Acquisition Conference, Rocky Hill, Connec ticut, Septem ber 1994. Inst i t ute o f Transportation Engineers. "Manual of Transportation Engineering Studies." Pre ntic e Hall, Englewood Cliffs, New Jersey, 1994. JHK &Associates, Inc. Hill sborough County Congestion wlanagernent System Work Plan." Sept ember 1994. Liu, T. ''Fie ld Tests of Travel Time Surve y Methodologies and Devel opment o f a Standardized Data Processing and Reporting System. National Traffic D ata Acquisition Conference, Rocky Hill, Connecticut, Septemb e r 1994. Manges, K., et.al. "Compa rison of Home Travel Survey Collection Methodologies." Transportation Planning Methods Applications-T hird National Conference, D alla s Texas April1991. McKinstry, D. and Nu ngesser, L. "An Eval uation of O n-Site Administered Origin-Destina t io n Survey Methodologies: Postca rd Mail-Back vs. Interv i ew." Third National Conf e re nce on Transporta t ion Planning Methods Appl icatio ns, Dallas Texas, April1991. Michal opo ulos, Panos G., et al. "Field Implementa t ion an d Testing of a Machine Vision Based Incident De t ection System. Proceedings o f 1st International Vehicle Navigation and Information Systems Conference, 1990. Miller, K., e t al. Using Video Technology to Conduct 1991 Boston Region Extern a l Cordon Survey." Transportation Research Record #1412, Transportation Researc h Board Washington D.C 1993. T'Y
PAGE 68

Demfmstration of Video-Based Tecbnology for Automation of Traffic Data Collec tio n Muntean, D "OriginDestinat i on T ravel Surveys in t h e 1990's U sing Microcomp u ters. Ins t itute o f Transportat i on Engineers, F lorida Section, 1993. Peterson, E ., and Hamb urg, J. "Travel Surveys: Current Options." Tra nsport a t ion Research Record #1097, Transpo rtati on Research Board, Washington, D.C. 1986. P i etr zyk M.C. An Application of ITS Technology for Congest i o n Management Systems." 65th Annual ITE M eet ing, Denver C o lorad o AuguStl995. Proposal f rom Hu ntingdon Engineering & Enviro nmental, Inc. (now T ransformat ion Systems, Inc ). "NHS Field D emonstration Project: Automated Video-Based T rafficData Collection. Hou ston, Texas, J anuar y 1995. Schaef e r M "License P late Match ing Suneys: Pract ical Issues and StatiStical Consideration." ITEjostrnal, Volume 58, No. 7,j uly 1988. Shuldiner, P aul \Y/. "Applications of Video and Machine Vision Technologies to Traffic Safety and T ransportat ion P lanni n g in the Uni ted States and Great Britain. Presented at the Strategic H i ghway Research Program and Traffic Safety Co nference, Prague, Czechoslovakia, Septem ber 20-22, 1995. Shul diner Paul W., e t al. "Determ inin g Detailed Origin-Dest i nation and Trave l T im e Patterns Us ing Video and Machine V i sion L icense Plate Matching." Presented a t the Annu a l M e eting o f the Transportation Research Board, Washington, D .C., January 1996. Turner S hawn, M., et al. Cong e stion Management: Data Collec t ion Methodologies." Texas Transpo r tation Instit u te> 1994. Turner, Shawn M. Advanced T echn iques for Travel Time Dat a Collect ion ." Proceedings of the 6t h International Vehicle Navigat ion &Informat ion a l Systems Con ferenc.e, Seattle, Washi ngton, Jul y 30-August 2 1995. Youn gb l ood W. "Integrat ion of Tr affic Operations and Traffic Data Collect ion." Nationa l Traffic Dat a Acquisition Conference, Rocky Hill, Con necticut, S eptember 1994. Zakaria, T. "Cordon Line Highway Survey for the Dela'l\are Valley Region." Transporta tion Research Record #1305, Tra nsportation Research Board, Washington, D.C 1991. 66 -=========== Trttot-l Origin-Destination. and VeiJicle Occupancy

PAGE 69

Demonstration of Video-Based Technology for Automation of Traffic Data Collection Project Contacts: Michael C. PietrLyk, P .E., Principal Investigator Senior Research Associate & ITS Progrom Manager Center for Urban Tronsportation Research College of Engineering, University of South Florida 4202 E. Fowler Avenue, ENB 118, Tampa, Florida 33620.535 Tel: (813)-974815 Fax: (813)-974-5168 e-mail: pictrzyk@eng.usf.edu Frank Kalpakis Principal Planner Hillsborough County Metropolitan Planning Organization 601 E. Kennedy Blvd., 18th Floor, Tampa, Florida 33602 -5117 Tel: ( 813 )-2n-5940 Fax: ( 813)-27U258 Jeffrey B. Woodson Presiden t Tronsfomation Syst ems, Inc. 2537 Sou t h Gessner, Suite 212, Houston, Texas 77063 Tel: (713)-9527494 Fax: (713)-7497 Salvatore A D' Agostino Presid ent Computer Recognition Systems, Inc. 639 Massac husetts Avenue, Cambridge, Massachusetts 02139 Tel: (617)-491-7665 Fax: (617)-491-7753 Pau l W. Shuldiner, Ph.D., P .E. Professor University of NWsathuwu 2148 Marnon HaU, Amherst, Massachusetts 0 1003 5205 Tel: (U3 }54S.2688 Fu: (413)-54588 Tra
PAGE 70

68 Demonstration of Video-Based Ttcbnowgy for Automatio n ofTr4fic Data Co/kct1on Center for Urboo T1"2nsportation Researth College of Engineering Univer>ity of South Florida 4202 E. Fowler Avenue, ENB 118, Tamp, FL 33620-5350 (813)-974-3120, Fax' (8U)-974168 email brosch@outr.eng.usf.edu web sit., httpd /www.cutr.cng.usf.edu/CUTR/ cutrhome.html Gary L. Brosch, Dirnto ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;= Trawl Timt, Origin-Dt-stination, and Aw.-Agt Vthi