基于HSV色空间的车牌定位综合方法

An effective Integration method for license plate location based on HSV color space

提出一种基于HSV色空间的车牌定位的综合方法,该方法的主要优点是避开了以往车牌定位方法中无法寻到合适的全局阈值的缺点。在HSV色空间内,构造5级灰度图,直接根据车牌颜色的不同来进行灰度化处理,再利用灰度形态学平滑方法进行消噪,利用字频统计等方法进行分析和判断,最后用投影法分割出汽车牌照。该方法充分利用牌照的底色、纹理、以及几何特征等,实验表明该方法具有较好的精确度,并且在不同的背景下定位的有效性。

An effective integration method for license plate location is presented based on HSV color space, whose advantage is to be able to determine the global threadhold. In the HSV color spce, we constuct a five-scale gray image for gray processing through various color of the license plate, use morphological method to eliminate noise, employ the statistical-letter-frequency property to analysis and judge the location, and then use projection method to separate the license plate. The proposed method can make full use of background, veins and geometrical features, and the experiments show that method has a high accuracy, and is effective for different background.