基于深度的3D视频质量评价  

Depth perception-based 3D video quality assessment

作　　者：孙延[1] 岑峰[1] 

机构地区：[1]同济大学电子与信息工程学院,上海201804

出　　处：《微型机与应用》2013年第3期45-48,共4页Microcomputer & Its Applications

基　　金：国家自然科学基金(60972035)

摘　　要：针对3D视频质量评价,基于人眼视觉系统HVS提出了一种新的加权SSIM评测方法。利用MVD深度视图空间和时间上的特性,提取主观权值映射,基于SSIM利用权值映射便得到了基于深度的SSIM加权算法DSSIM。实验结果表明,DSSIM比PSNR更趋近于HVS,比SSIM更充分地考虑了3D结构信息,得到更准确的3D视频质量评价。This paper proposes a new weighting SSIM evaluation method based on human visual system（HVS） for the 3D video quality assessment. The opinion weighting map is extracted from the features of depth view on space and time of the multi-view video plus depth（MVD）. The depth-weighting SSIM（DSSIM） is proposed to assess the quality of 3D video, taking the advantage of weighting map based on SSIM. Experimental results show that, the proposed DSSIM has a better performance in 3D video synthesis compared with PSNR. This algorithm is also better than SSIM, because of the full consideration of structure information of 3D Video.

关 键 词：3D视频质量评价 深度视图 权值映射 SSIM 3D视频合成 

分 类 号：TN919.8[电子电信—通信与信息系统]