本程序采用一种基于垂直Sobel算子检测边缘和投影法的车牌定位方法,根据车牌区域竖直纹理突出的特点，利用Sobel垂直算子提取边缘，然后投影得到车牌的水平投影图，利用水平投影图分割出车牌，再运用膨胀运算进行车牌垂直定位。利用旋转投影法寻找车牌倾斜角度，然后用双线性插值进行车牌图像倾斜矫正。对经过精定位的车牌利用垂直投影法，找出各个字符的中心点，根据字符中心点位置进行切割。同时将切割出来的图片与数据库中的每张模板图片进行减法运算，计算该图片与模板的均方误差，其均方误差最小的模板即为识别出的字符。-This program uses edge detection and projection of license plate positioning method based on vertical Sobel operator, according to a prominent feature of the license plate area vertical texture operator Sobel vertical edge extraction, then the horizontal projection of the projector to get the license plate, the horizontal projection the Figure split out the license plate, and then use the the dilation operation license plate vertical positioning. Take advantage of the rotation projection method to find the license plate tilt angle, and then the license plate image tilt correction bilinear interpolation. The focal point for fine positioning license plate using the vertical projection method to identify each character, according to the the character center point cut. Cutting out pictures with each template images in the database, subtraction, calculation of the mean square error of the picture with the template, the mean square error of the smallest template is identified characters.