摘要
轮胎印痕图像增强是进行车辆胎痕图像特征点提取及匹配最为关键的预处理阶段。以一次事故中现场拍摄衣物上的轮胎痕迹为研究对象,针对一些传统的图像增强方法的不足,提出了基于多尺度Retinex轮胎痕迹增强方法。为了进行对比,实验中还引入单尺度Retinex算法。实验结果表明:算法能提高胎痕图像对比度且能显著增强图像细节,且较单尺度的Retinex算法,在实现压缩动态范围的同时具有较高的色彩保真度,得到了期望的图像效果。
Enhancement of tire prints image is the most critical preprocessing stage before feature extraction and points matching.The tire marks picture on an accident is regarded as the research object.According to some deficiencies of traditional methods in image enhancement,multiscale Retinex method of tire prints enhancement is proposed.In the experiment,the singlescale Retinex is introduced in order to compare the enhancement results with MSR＇s.As a result,the experiment show that the MSR algorithm can enhance the picture contrast together with details,but also can has a high color fidelity,dynamic range compression and so on.So the desired graph is obtained.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第30期192-194,共3页
Computer Engineering and Applications
基金
国家自然科学基金No.60803088
陕西省自然科学基础研究计划项目(No.2009JM8018)
中央高校基本科研业务费专项资金重点项目(No.GK200901006)~~
