基于网格的立体图像编码算法  

Stereo image coding algorithm based on mesh

作　　者：杨嘉琛[1] 侯春萍[1] 雷建军[1] 张卓筠[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《光电子．激光》2010年第1期136-139,共4页Journal of Optoelectronics·Laser

基　　金：国家"863"计划资助项目(2007AA01Z304);国家自然科学基金资助项目(60672077)

摘　　要：基于特征点视差估计与三角网格映射,提出一种立体图像压缩编码算法。为了保持压缩后视点对的视差不发生变化,利用绝对差值图进行特征点选取。进行残差图像编码时,结合了心理立体视觉影响、人眼亮度色度特性、立体图像对色度特点与三角网格映射特点等因素,提出了立体残差编码只需对Y分量进行的方法。实验表明,该算法具有一定的优越性,如压缩比大,图像质量与立体感均较好。A stereo image coding algorithm is proposed based on disparity estimation of feature points and triangle mesh mapping.To maintain the disparity of two view-points after coding,the feather points with the absolute disparity image are extracted.When coding from the residual pictures,the stereo residual coding is conducted only in Y vector based on the influence of the psychological stereo perceive,luminance and tint feature of human eyes,the tint feature of the stereo image pair and feature of the triangle mesh mapping etc.The experiments demonstrate that the proposed algorithm has larger compression ratios,and the quality and stereo feeling of the decoded stereo image are better compared with JPEG and other algorithms.

关 键 词：立体图像 图像编码 网格 视差估计 

分 类 号：TN91[电子电信—通信与信息系统]