基于数学形态学的车牌定位研究与实现

Car License Plate Location Research and Implementation Based on Mathematical Morphology

车牌定位是车牌识别系统中的关键环节,定位的准确率是衡量系统性能的重要技术指标。针对预处理后的灰度图像,利用了数学形态学的腐浊、膨胀以及开、闭等运算,消除无用信息和噪音,然后计算出图像中连通区域的面积,并根据图像的实际情况确定面积阈值。研究并提出了基于形态学综合考虑车牌特征的算法。给出了从图像预处理到车牌定位的完整实现过程及结果。实验分析表明该算法克服了单一特征算法因不完备而引起识别率低的缺点,具有良好的定位效果。

Car license plate location plays a key role in car license recognition system.moreover,positioning accuracy is a critical technical indicator to measure system performance.And after preprocessing,in the gray scale image the useless information and noises have been eliminated by erosion of mathematical morphology,dilation as well as opening and closing operation etc..Then the area of the connected component is computed.Based on the consideration of both features of car license plate and morphology,this study proposes an algorithm which offers a complete and actual process and results from image processing to car license plate location.And the experimental analysis shows that this algorithm can overcome the shortcoming of low recognition rate caused by unimproved single feature algorithm and in effect the positioning performs eminently.