Car plate recognition system based on chain code and local signature

Abstract

Car plate recognition system has been developed to identify vehicles by the contents of their car plate for law enforcement nowadays. The application of the recognition system has wide variety where it is used as the vehicle access control, state border control and vehicle traffic monitoring. Since the usage model is wide, the accuracy of the recognition system is very important. This project is aimed for implementation of a car plate recognition system for standard Malaysia car plates using the chain code and local signature as the recognition technique. This system consists of several phases which are the car plate image extraction, car plate characters segmentation, database creation and character recognition. Besides, Graphical User Interface is developed for better user experience. The software is developed in the Visual C++ environment and the image processing is done by the OpenCV library. With the 100 images that being tested, the success rate for the car plate image extraction is 94%, segmentation is 93% while recognition is 75%. This recognition system does not give high recognition accuracy but improvement is suggested on the software development such as adaptive threshold mechanism and chain code normalization prior to recognition phase.