基于边缘统计和颜色特征的车牌综合自动定位方法

车牌识别在智能交通系统中起着重要作用.车牌定位是车牌识别中的关键步骤.本文提出一种基于车牌字符边缘统计和颜色特征的综合定位方法,可以有效地解决背景复杂的彩色图像中车牌定位的问题.该方法分为竖直边缘检测、边缘统计分析、车牌候选区定位、候选区筛选、车牌倾斜矫正.通过对垂直边缘的统计分析将邻近的边缘点进行连接,结合车牌的位置、颜色等特征对连接形成的块状区域进行筛选,而后对得到的车牌区域加以校正,最终输出易于分割的车牌字符图像.该系统包括从图像采集,到车牌分类、车牌文字区别等完整过程,适应性强.通过一系列实际采样图像的试验结果证明,该方法准确率高、鲁棒性好,能够满足实际车辆车牌自动识别系统应用的需要.

License plate location based on edge statistics and color features

LPR (Automatic License Plate Recognition) plays an important role in numerous applications. License plate location is the key component of LPR system. A hybrid license plate location method based on edge statistics and color feature is presented. The method can efficiently cope with the color images of complex background. The proposed algorithm can be divided into five sections, which are vertical edge detection, edge statistical analysis, candidate regions location, candidate regions filtration, tilt correction. We connect the neighboring edge points by the statistical analysis method and pick up the rectangular region of the LP by natural characteristics such as position, color features, etc. Afterwards, a process of correction should be added to. The proposed method proved highly accurate and robust through the experiments with a set of practical sample images. It also provided some new algorithms about correction of the errant character locations. At the end of the thesis we set examples about our identification result to certify the proficiency of the whole AVI(Automatics vehicle Identification )system.