摘要
车牌定位作为车牌识别系统中的基础环节,定位准确度将直接影响着最终识别结果。在以往各种识别算法中,Hough变换更多的是用来检测车牌倾斜角度,在车牌定位中一直未见其用。结合合理的先验知识,提出一种利用Hough变换多线检测在车牌定位中的新用法。通过对240幅图像的测试,准确率达到95.42%,结果表明了此方法的有效性。
License plate locating is the base link of license plate recognition （LPR）, so the accuracy has a direct influence to the finally recognition result. In various algorithms existed, Hough transformation is usually used on detecting plate angle,rarely be used on locating. A new method of plate locating by multi-line defection based on Hough transformation is presented, which is combined with prior knowledge. The locating accuracy rate of 240 images achieves 95.42%, the results show that it is an effective method.
出处
《计算机与数字工程》
2009年第8期138-140,共3页
Computer & Digital Engineering
基金
湖南省教育厅科学研究项目(编号:06C519)资助
关键词
先验知识
HOUGH变换
多线检测
prior knowledge, Hough transformation, multi-lines defection
