结合跳变法和SVM的车牌精确定位方法

Method of Accurate Location of License Plate with Combination of Jump Method and SVM

针对车牌的精确定位,首先采用边缘检测和形态学处理相结的方法,依次对目标图像进行灰度化处理、高斯模糊处理、Sobel运算、二值化、闭操作、取轮廓处理,初步筛选出符合条件的候选区域,再利用车牌字符与字符间的间距特征,使用跳变法对候选区域进行进一步筛选、经过SVM分类器判定,得到最终的车牌区域,为后续的车牌字符识别作准备。经验证,提出的车牌定位算法快速准确,对环境有较好的适应性,具有一定的实用性。

According to the accurate location of the license plate,this paper adopts the method of edge detection and morphological processing combined,in order to target gray image processing,Gaussian Blur,Sobel algorithm,binarization,closed operation,contour processing,screening out the candidate region to meet the conditions,the spacing between characters and character the use of jump method are further screening of the candidate region.Through the SVM classifier to determine,get the final license plate area,preparing for the subsequent character recognition of license plate.It is proved that the license plate location algorithm proposed in this paper is fast and accurate,has good adaptability to the environment and has certain practicability.