Virtual View Synthesis Using Generated Global Background

Virtual view synthesis is the key technology for many applications such as 3D-TV and multi-view video. As an effective and mature technology, Depth Image Based Rendering (DIBR) has obtained much attention recently. However, DIBR also brings a new problem, namely disocclusion. Accordingly, how to reduce the disocclusion exposed during the projecting process has become one of the major challenges in DIBR. To solve this problem and improve the quality of synthesis image, as well as decrease the number of reference views which are desired during the synthesis process, a novel algorithm, namely, virtual view synthesis based on global background, is proposed in this paper. Firstly, the inter-frames correlation information in the video sequence is utilized to generate a global background image. Then the generated global background image would be treated as an auxiliary image to complement the images which are projected to the virtual view. In contrast to the existed approaches, the proposed algorithm fills the disocclusion areas with their true pixel value, which brings a higher image quality. Finally, the inpainting technology is applied to repair the remaining exposed regions. The experimental results indicate that the proposed algorithm performs better than the traditional ones. Moreover, the presented approach only depends on one reference view.