摘要
随着彩色摄像机广泛使用,对彩色图像的视差匹配研究越来越普遍。研究显示,使用彩色图像匹配比灰度图像匹配精确度增加20%～25%。根据非参量变换原理,提出了基于非参量变换的彩色图像立体匹配算法,将Rank变换应用到彩色图像的区域匹配方法中,并采用census变换后的汉明码距离作为新的约束条件。实验表明提高了原来方法的抗噪性和匹配准确性。
Research of color image stereo matching becomes along with wide use of color camera. It is shown that the precision of matching results are always improved by 20% to 25% when using color information instead of gray value information. Presents color image matching approach based on non-parametric transform. With non-parametric theory, applies rank transform to color image local area matching. Hamming distance between pixels after the census transform is presented as a new constraint. The experiment result indicates that the proposed algorithm is the more precise matching and invariance to certain types of image noise than the ones based on intensity.
出处
《计算机科学》
CSCD
北大核心
2008年第11期217-219,共3页
Computer Science
基金
国防基础研究项目(编号:J1500C002)基金资助
