Vehicle License Plate Localization and Recognition System for Intelligent Transportation Applications

Traffic surveillance systems play an important role in the improvement of traffic safety. Several traffic problems need to be controlled, due to the huge number of vehicles that increases daily, such as traffic accidents, the theft of vehicles and traffic violation. License Plate detection and recognition system (LPDRs) in natural scene images is one of the challenging systems in traffic surveillance. In this paper the LPDRs is the main goal. The proposed method follows two principal steps: hypothesis generation step and hypothesis verification step. In the first step, an adaptive threshold method is used to cope with the dynamic changes of illumination conditions during the binarization of the image. Therefore, the otsu's algorithm is applied which is an efficient and simple method. Followed by performing the Connected Component Analysis technique (CCAT) to detect the rectangles which are the generated license plate candidates. In the second step, the edge detection is applied inside the generated candidates then the close curves method is performed to ensure that the candidate is a license plate and to segment the character. Followed by performing the cross-correlation method between the segmented characters and the templates to define the license plate characters. The proposed method has a good performance in term of robustness where the experiment results show that our system achieves a good accuracy.