Real-time recognition of license plates of moving vehicles in Sri Lanka

Abstract

Automated license plate recognition can be used for many applications such as detecting traffic light violations, access controlling, calculating parking fee and so on. However, detection and recognition of license plates can be seen as a complex problem. This paper presents an algorithm which can be used in Sri Lanka, for detecting and recognizing license plates automatically using image processing and neural networks techniques. In the proposed algorithm, the license plates are located by analyzing the regions with highest vertical edge density. Hough transformation and the affine transformation techniques are used to handle the skewed license plates. After extracting the license plate characters, a neural network is used to recognize those characters. The experimental results show that the proposed system can successfully detect and recognize all types of license plates in Sri Lanka and is suitable for real time implementation because of the lower execution time.