A license plate recognition system and a new license plate location algorithm

which is called Vector Map Location Algorithm (VMLA)

are introduced in this paper. The algorithm is based on the feature that the stroke of license plate characters and numbers have two edges which have the maximum cross correlation values. Thus by using the same method of flow PIV (Particle Image Velocimetry)

a kind of vector map can be calculated and showed. VMLA is a real time and plate size adaptivity algorithm at a large range. At the same time

the contrast condition about the character to the background and a coarse estimation to the width of the stroke of license plate characters and numbers also can be derived from the map. Those above information are important to location precisely

character segmentation and recognition. The testing result of 540 images is that it costs about 0 2 second to process an image and 98 9 percent of the picture have good result

which shows that VMLA is a fast

high accuracy rate and promising location algorithm.