Automatic vehicle license plate recognition system used in expressway toll collection

Based on the binary image, the thesis proposes a suit of algorithm for the license plate intelligent recognition system serving the expressway toll collection system. Firstly, the algorithm of first difference combined with the method of determining the threshold of binarization via accumulating pixel spots is used in pre-processing. Secondly, taking advantage of the horizontal texture characteristic and the physical trait in the plate area, the algorithm locates the position of the plate. Then the positions of left and right border of Chinese character used to division are marked in character segmentation, so that the strokes of Chinese characters cannot degenerate noise. At the end, the algorithm designs different methods of recognition, the enhanced method of BP neural network is for the Chinese characters, and the method of dynamic clustering for Arabic numbers and alphabet letters.