基于HSV与边缘信息的车牌定位与字符分割方法

License Plate Location and Character Segmentation Based on HSV and Edge Information

研究了一种车牌定位和字符分割方法:采用HSV颜色空间与边缘信息结合的方式进行车牌定位,利用垂直投影判断字符界线的方式进行字符分割。首先将原始图片转换到HSV空间,然后利用车牌的颜色信息粗定位车牌,最后根据已知的车牌比例确定车牌具体位置;对于倾斜的车牌,利用仿射变换来校正。将定位得到的车牌提取出来,进行字符分割预处理;然后统计预处理后车牌的像素值所在位置,存储为矩阵,并且画出垂直直方图,在垂直直方图中可以明显看到每一个字符的边界位置,将此边界作为依据进行字符分割。经实际验证,上述车牌定位和字符分割方法准确可靠。

This work studied a method of license plate location and character segmentation:license plate is positioned by combining HSV color space and edge information,while the characters are segmented by vertical projection method of determining the character boundary.First,the original picture is converted into HSV space,and then the license plate is roughly located by the color information.Finally the specific position is determined according to the known proportion of the license plate.For slanted license plates,affine transformation is used to make correction.After pretreatment,the pixel value of the license plate is stored as a matrix and the vertical histogram is drawn.The boundary position of each character can be clearly seen in the vertical histogram.The boundary is taken as the basis for character segmentation.The above method is verified to be accurate and reliable.