基于互相关矢量图的车牌定位新算法

互相关矢量图是由流场粒子图像测速的互相关计算结果绘成的矢量图形,提出一种车牌自动定位的新算法--VMLA算法,该算法基于字符笔画左右边缘之间的互相关矢量图的结构特点,可以从复杂车牌图像背景中搜索到车牌的正确位置.VMLA算法具有较强的车牌大小自适应性和实时处理能力,540幅车牌图像的实验结果初步验证了该方法的有效性.

New location algorithm of car license plate using cross-correlation vector map

Vector map is a kind of computer graphics formed by matrix of vectors. These vectors are results of cross-correlation of two related particle images. This technique is called PIV(particle image velocimetry). A new algorithm for extraction of car license plate is introduced, which is called VMLA(vector map location algorithm). The idea of VMLA is inspired by the method to get the cross-correlation vector map of PIV. It based on the fact that the two edges of strokes of license plate characters and numbers cause a special local structural pattern in the cross-correlation vector map. Using this idea, the correct position of a plate could be extracted from vector maps of license plate images with complicated backgrounds. VMLA is adaptive to various license plate sizes and it also runs in real-time. Testing results on 540 license plate images show that the method is valid and efficient.