Reduction of Video License Plate Data

Abstract

Video license plate surveys have been used for more than a decade in Indiana to help produce origin-destination tables in corridors and small areas. In video license plate surveys, license plate images are captured on videotape for data reduction at the analyst’s office. In most cases, the letters and numbers on a license plate are manually transcribed to a data file. This manual process is tedious, time-consuming, and expensive. Although automated license plate readers are being implemented with success elsewhere, their dependence on high-end equipment makes them too expensive for most applications in Indiana. Presented are the results of an attempt to use standard video cameras and tapes, readily available video processing equipment, and open-source software to minimize the human role in the data reduction process and thus reduce the expenses involved. The process of automatically transcribing video data can be divided into subprocesses. Analog video data are digitized and stored on a computer hard disk. The resulting digital images are further processed, by using image-processing algorithms, to locate and extract the license plate and time stamp information. Character recognition techniques can then be applied to read the license plate number into an electronic file for the desired analysis. The described video license plate data reduction (VLPDR) software can identify video frames that contain vehicles and discard the remaining frames. VLPDR can locate and read the time stamps in most of these frames. Although VLPDR cannot read the license plate numbers into a data file, this final step is made easier by a user-friendly graphical user interface. VLPDR saves a significant amount of manual data reduction. The amount of labor saved depends on the parameters chosen by the user.