摘要
提出了基于数学形态学和纹理特征的快速定位车牌的方法.该方法包含4个步骤:预处理,利用Sobel算子进行垂直边缘检测,通过形态学操作得到连通域,利用车牌的纹理特征和字符特征最终定位车牌.使用该方法对1 144幅图片进行实测,1 102幅图片定位成功,正确识别率为96.3%,处理时间为160~200ms.
Based on mathematical morphology and texture features, a practical and fast method for vehicle license plate reeognition was presented. The method consisted of four steps: preprocessing, vertical edge detection using Sobel operator, morphological operation for obtaining connected components and plate recognition using texture features. In the experiment, 1 102 among 1 144 images were successfully located, corresponding to a recognition rate of 96.3%. The recognition time for one plate was 0. 16--0.20 ms.
出处
《物理实验》
2012年第6期13-16,共4页
Physics Experimentation
关键词
SOBEL算子
数学形态学
结构元素
连通域
纹理特征
Sobel operator
mathematical morphology
structural element
connected region
texture {eature