By: Haluk Laman, Amr Oloufa
Summarized Description:
In order to collect O-D data that includes the start points, end points, and times of truck trips, a research project with three phases worth $600K in total funded by FDOT District 5 was conducted during the time period from January 2012 to November 2017. The aim of this research project was to develop a novel method of automated real time O–D data collection that is reliable, inexpensive, and portable using a mix of commercial off-the-shelf hardware and custom software. As such, the researchers conducted an automated license plate reading methodology. The first step was to identify the length of highway in which cameras could be installed such that license plates would be in view and three stations could be set up to get the maximum interpretation of origins and destinations. The second step included selection of the appropriate hardware configuration (i.e., camera and trigger systems) and a solar power design for each location that would be cost effective and safe. The third step was the installation of the hardware and solar power components. The last step was the development of software to process and interpret the collected O–D data.
The plates were processed by optical character recognition (OCR) software and the results of the OCR would be stored in a database. The results were then analyzed using database and pattern-matching techniques to show when a truck entered and left this small transportation network. The researchers utilized a three-step approach to match license plate reads between locations. The first attempt to match was a simple database query based on exact matches within a selected time frame. The second attempt was to use a fuzzy match using database queries with wildcards and substitutions for common OCR misreads. The third attempt used partial matches via coded algorithms. Any reads that were unmatched either were assumed to have entered or did enter the network between locations.