基于车牌特征和底色识别的车牌定位算法研究

Research on Location Algorithm Based on Vehicle License Plate's Structure Feature and Grounding-color Recognition

在车牌识别系统中,车牌定位是车牌自动识别系统中最重要的一步,车牌定位的准确程度直接影响车牌识别的精度。针对背景复杂,车牌区域模糊的图像,提出一种车牌结构特征和底色相结合的车牌定位方法。该算法首先利用彩色边缘算子提取图像的边缘,然后利用连通域算法找出不同的连通区域,最后结合车牌的结构特征(宽高比、车牌区域灰度跳变次数)和车牌底色特征(目前是黄色、黑色、白色和蓝色)选定最佳区域。实验结果表明,该算法简单,定位准确,满足实时性要求。

In the license plate recognition system,license plate localization is the most important step of the automatic license plate recognition system.The accuracy of License plate localization directly affect the accuracy of license plate recognition system.According to the complicated background and license plate area fuzzy image,this paper presents a combination method of license plate structure characteristics and grounding-color.This algorithm firstly using color edge operator extract the edge of the image,and by using the connected domain algorithm to find different connected area.At last,combined with the features of the structure of the license plate（aspect ratio,jump frequency of license plate area gray image） and grounding-color（At present is yellow,black,white,and blue） selected the best area.