ALPR stands for "Automatic License Plate Recognition".
When a license plate is captured by the video camera, a trigger is caused which initiates a suite of software recognition algorithms. These algorithms identify the position of the plate in the image, extract the plate image, apply geometric distortion compensation if necessary and segment the plate into individual characters. Each character is then fed into a software neural network. The network produces an estimate for every character input resulting in a string of characters, which is its best estimate for the license plate. Multiple software based processing units, provide multiple parallel processing ensuring that many images will be analyzed for each passing vehicle. This, in turn, results in the extraordinary accuracy of the PlateScan system. The PlateScan processor unit then selects the best result from all the results generated by the software processors for each passing vehicle. This is made possible by the neural network assigning a confidence score to each character it recognizes.
 PlateScan in daily use
PlateScan’s proprietary software is supplied installed on the PlateScan trunk mounted computer, which, in turn, is connected to several vehicle mounted cameras optimized for license plate recognition. Every license plate scanned is compared to a list of “vehicles of interest” associated with auto theft, felony warrants, BOLOs, parking violations, or any other license plate-oriented databases, such as Amber Alerts, DOJ & NCIC downloads, CAD/RMS data extracts, etc., the police agency wishes to use. PlateScan’s proprietary software is supplied installed on the PlateScan trunk mounted computer, which, in turn, is connected to several vehicle mounted cameras optimized for license plate recognition. Every license plate scanned is compared to a list of “vehicles of interest” associated with auto theft, felony warrants, BOLOs, parking violations, or any other license plate-oriented databases, such as Amber Alerts, DOJ & NCIC downloads, CAD/RMS data extracts, etc., the police agency wishes to use.
PlateScan’s proprietary software is supplied installed on the PlateScan trunk mounted computer, which, in turn, is connected to several vehicle mounted cameras optimized for license plate recognition. Every license plate scanned is compared to a list of “vehicles of interest” associated with auto theft, felony warrants, BOLOs, parking violations, or any other license plate-oriented databases, such as Amber Alerts, DOJ & NCIC downloads, CAD/RMS data extracts, etc., the police agency wishes to use. When a positive match is received, officers are alerted by an audio and/or visual alarm. They, then, respond to the “hit” in the appropriate manner for that agency.
Until an alert occurs, the PlateScan system operates in the background, allowing them to perform their regular police activities with full integration to existing MDT systems and CAD/RMS software. PlateScan software uses an intelligent Neural Network based recognition engine, providing a distinct advantage over base-level solutions, such as OCR, as it significantly enhances the accuracy rate. In addition, PlateScan’s Neural Network features distributed parallel processing. This means that it is capable of accurate high speed plate recognition across multiple lanes of high density traffic. The system is currently capable of processing 4 vehicle license plates per second and reading 4 recognition cameras simultaneously, each with separate overview images. Research has shown that patrol officers equipped with the technology can have arrest rates significantly higher than officers working without it. This will deliver reductions in crime, enhanced community safety and safer roads. |
