自然复杂环境中基于颜色的多车牌定位研究

Multi-License Plate Extraction Based on Color Features in Nature Complex Environment

为解决自然复杂环境中车牌的定位和识别问题,提出了一个结合车牌颜色特征、灰度特征、几何形状特征的通用车牌定位方法,并首先根据中国车牌的颜色特征,提出了一种基于距离的颜色分类方法,解决了颜色分类的基本难题;然后采用行扫描及线段分析的方法得到车牌底色所在的矩形区域;最后将可能的车牌区域统一转换成底色为黑色,字符颜色为白色的二值图像,再利用灰度特征进行验证,并分割出车牌区域。采用自然复杂环境中拍摄的图像对新方法进行验证的实验表明,与其他方法相比,新方法可在一定限度内自适应车牌的类型、大小、数量和方向,并对汽车在图像中的位置以及图像背景的限制较少,是一种适用性较强的方法。

To address the License Plates （LP） extraction in complex natural environments, a novel approach to LP extraction was presented, which utilized LP＇ s color information, gray feature and geometry characteristics. Firstly, a color classification method was proposed based on the distance between different colors according to the Chinese car license plate color features. Secondly, LP＇ s background color zones were segmented by scanning lines of picture and analyzing the line segments. Finally, these Zones were translated into binary image in which LP＇ s background color was dark and LP＇ s foreground color was white, and validated by the LP＇ s gray features. In a certain degree, the method could adapt to LP＇ s type, size, number and direction and wash＇ t limited to the location of car the and the background in the picture. It was tested using the natural pictures, and was approved to be effective and adaptive.