基于投影特征曲线匹配的车牌字符识别算法  

A Recognition Algorithm for License Plate Recognition Based on Projection Characteristics Matching

作　　者：姜瑾[1] 张桂林[1] 许慧慧[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,武汉430074

出　　处：《计算机与数字工程》2007年第6期20-22,91,共4页Computer & Digital Engineering

摘　　要：在车辆牌照自动识别系统中,字符识别是最后一个也是最关键的步骤。在实际场景中,光照不同造成了车牌字符效果不同,从而影响字符识别的正确率,尤其是车牌中汉字的识别。提出一种灰度预处理算法来解决识别过程中的光照问题。针对不同字符在垂直方向和水平方向投影特征曲线不同,采用分类比较和滑动匹配方法来识别车牌字符,提高算法的适应能力。Character recognition is a last and key process in the vehicle license plate recognition system. In the real scene, illumination problem often impact the discrimination rate,especially the recognition of the chinese characters. In this paper, an arithmetic uses the gray scale and the projection feature of the character is proposed. The method of the gray scale’s pretreatment is slected as stage one for solving the illumination problem. Then the different projection characteristics in the horizontal and vertical direction are using to identify character to be as stage two.Also we use sort comparison and slide matching to boost the adapt capacity of the arithmetic.

关 键 词：字符识别 灰度预处理 投影特征曲线 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]