用于多光照条件下车牌定位的遗传算法  

LOCATING LICENSE PLATE UNDER VARIOUS ILLUMINATION CONDITIONS USING GENETIC ALGORITHM

作　　者：熊军[1] 高敦堂[1] 都思丹[1] 沈庆宏[1] 

机构地区：[1]南京大学电子科学与工程系,江苏南京210093

出　　处：《计算机应用与软件》2005年第12期107-109,共3页Computer Applications and Software

摘　　要：针对多光照条件下由于车牌图像退化难以定位的问题,提出一种基于遗传算法的解决方案。该方法首先对图像局部区域进行二值纹理特征增强处理,并通过一组一维滤波器获取其纹理特征向量,最后使用基于实整数编码的遗传算法对全图进行车牌特征匹配搜索最终准确定位车牌。测试结果表明,结合遗传算法的局域特征增强算法对多种光照条件下的抓拍图像定位性能稳定,定位成功率优于传统的直线检测方法和灰度变化特征检测方法。In this paper genetic algorithm is used for the license plate locating problem under various illumination conditions. At first, a possible local area of input vehicle image is processed to enhance the binary texture features and the feature vector is retrieved from a bank of filter. Then the real integer coded genetic algorithm is proposed to search the whole image for a possible license plate. Test results has shown that the local area processing combined with GA is quite robust with images of various illumination conditions.

关 键 词：多种光照条件 车牌定位 遗传算法 图像增强 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U495[自动化与计算机技术—计算机科学与技术]