Depth image segmentation for improved virtual view image quality in 3-DTV

A new 3-D data representation format consisting of monoscopic color video and its associated per-pixel depth information is a most promising technique applied in future three-dimensional television system. Despite many advantages of this representation, a large problem which is so-called disocclusion still exists when synthesizing one or more virtual views of a real-world scene by means of depth image based rendering technique. In this paper, we first give an interpretation to the occurrence of disocclusion from a mathematical view, and then analyze the reasons why smoothing depth image can have some effects on disocclusion and its several drawbacks. Finally, we propose a global processing method using depth image segmentation to solve the disocclusion. Analytical and experiment results show that our proposed method can improve the quality of virtual view images to a large extent, and no geometric distortion arises.