基于颜色信息定位车牌及其校正

The Extraction of the License Plate Based on Color and Correction

车牌定位是车牌识别中的关键技术之一。本文对传统基于边缘检测的车牌定位算法做了改进,根据彩色图像R、G、B三者的分布特点,采用直接二值化定位车牌。在车牌区域提取之后,针对传统灰度化再二值化进行字符分隔对光线敏感的缺点,本文跳过灰度化过程,基于R颜色分量进行二值化,这种方法很好减少了阳光照射的影响。基于Hough变换,提出了基于双线性插值的两步法的校正方式。本文提出的方法运算过程较简便,代价低,可应用于嵌入式系统低成本硬件上,实现实时车牌识别。

The accurate extraction of the license plate is one of key steps for ANPR system. It makes improvement to the traditional method, locates the license plate directly according to the relation of the R, G and B. An improved method making binearization according to the value of R directly and skipping the process of graying is proposed in this paper. This method can reduce the impact of sunlight effectively. After that, by using Hough Transform and bi-linear interpolation, a two-step plate correction algorithm is presented. The proposed method and algorithm in this paper are more briefness and with lower cost, and they can be used in embedded systems of lower cost to recognize the vehicle plate.