基于OpenCV的车牌定位方法  

Mothed of vehicle license plate location based on OpenCV

作　　者：谭同德[1] 王三刚[1] 

机构地区：[1]郑州大学信息工程学院,河南郑州450001

出　　处：《计算机工程与设计》2013年第8期2816-2820,共5页Computer Engineering and Design

基　　金：河南省国际科技合作基金项目(104300510058);河南省自然科学基金项目(0611051900)

摘　　要：提出一种基于开源视觉库OpenCV从复杂环境中准确定位车牌的方法。利用车牌的颜色特征,将原图像分别在HSV颜色空间和RGB颜色空间下处理得到两幅二值图像;根据这两幅二值图像的纹理特点,在HSV颜色空间下得到的二值图像定位出车牌的上下边界,再按照定位出的上下边界坐标从RGB颜色空间下的二值图像上水平切割出车牌区域;根据车牌的几何特征,从切割出的水平区域中得到精确的车牌区域。实验结果表明,该方法能够快速、准确定位出车牌,还具有很强的抗干扰性。An accurate positioning license plate based on open source vision library OpenCV from complex environment is presented.Using color feature of license plate,the original image,is handled respectively,in the HSV color space and RGB color space processing to get two binary images.According to the texture features of two binary images,the upper and lower boundaries of the license plate from binary image are gotten in the HSV color space.The plate horizontal area is obtained by cutting the binary image on the RGB color space in accordance with the positioning out of the upper and lower boundary coordinates.According to the geometric features of license plate,accurate plate region from the cutting level area is obtained.The experimental results show that this method can quickly and accurately locate the plate,but also has a strong anti-interference.

关 键 词：开源计算机视觉库 车牌定位 HSV颜色空间 数学形态学 最大类间方差法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]