A Comparison Between the Use of CNN and Matching Templates in Recognizing the Iraqi License Plate Number

Abstract

One of the important applications that organize people's lives is the applications that relate to organizing the traffic regulation, detecting violators, detecting stolen cars, and managing car parks. Recognizing the license plate number (LPN) crosses the basic process in all previous applications. This process is affected by the surrounding light conditions while taking an image of the car. In addition to these problems, the plates of modern Iraqi cars (the so-called German number) suffer from words in silver color printed on the plate, which causes a senior problem when locating the LPN and recognizing it. To solve these problems, we presented in this research a comparison between the use of the SIFT and Contour algorithm in locating the LPN. While using CNN training models to recognize the PN achieved higher results than template matching.