融合字符纹理特征与RGB颜色特征的车牌定位

License Plate Location Based on Characters Texture Feature and RGB Color Feature

由于光照、距离变化以及背景复杂等原因,传统基于单一特征的车牌定位算法很难获得满意的效果。为完成复杂背景下车牌定位,提出融合字符纹理特征与RGB颜色特征的定位算法。利用车牌区域背景底色与字符颜色互补特性,通过寻找图像车牌区域颜色点对,结合字符纹理特征反馈校验,能快速准确定位车牌。对复杂背景下获取的12 581幅彩色图片进行实验,结果表明,定位准确率能够达到99.4%,验证了该算法的有效性。

Due to the change of illumination, distance and complicated background, etc. the desired effect can be hardly achieved by traditional license plate location based on single feature. In order to accomplish license plate location under complicated background, license plate localization algorithm which combines characters texture features with RGB color features has been proposed. The license plate can be located quickly and accurately by using the complementary features of bottom color and character color and then looking for color pairs on the license plate as well as combine the textural features which are used for feedback checking. The results show that the accuracy rate of license plate location can reach 99.4% by experiment on 12 581 color images under complicated background which has proved the validity of this algorithm.