3-D Video Quality Enhancement by Non-Linear Depth Map Configuration Based on Moving Objects
This paper proposes a new scheme of non-linear depth map configuration based on moving objects for multiview depth estimation in order to enhance 3-D video quality. Generally depth map has totally 256 depth levels as 8-bit intensity. The total range of depth map is divided into five groups with same space. By keeping track of position of moving objects in depth map, the group with higher movement is dynamically chosen. And the depth intensity intervals in the selected group are divided more densely. Consequently, this group has more depth levels to be represented in the same range of depth map and good visual effect. Experimental results show that the proposed scheme provides higher both video quality and peak signal-to-noise ratio (PSNR), whereas it adds low additional computational complexity.
3-D video quality Multi-view depth map rearrangement moving objects FTV view synthesis depth estimation
Tae-Woo Kim Jitae Shin
School of Information and Communication Engineering, Sungkyunkwan University Suwon, Korea
国际会议
2011 4th International Congress on Image and Signal Processing(第四届图像与信号处理国际学术会议 CISP 2011)
上海
英文
2611-2614
2011-10-15（万方平台首次上网日期，不代表论文的发表时间）