Virtual View Rendering Based on Self-adaptive Block Matching Disparity Estimation

Virtual view rendering is one of the key techniques of interactive 3D video systems, and is analogue expression to the real world in virtual reality systems. This paper proposes a new algorithm of virtual view rendering between two views taken in the same scene at the same time, which is based on self-adaptive block matching disparity estimation. This algorithm outputs any view between the two origin views. In order to gain accurate dense disparity field, a method of self-adaptive block matching is proposed in disparity estimation according to the information of texture and gray scale. Image interpolation is used to render certain virtual view and reverse mapping is adopted to fill up holes formed in interpolation. Satisfying visual effect is shown in experimental result.