An Efficient View Interpolation Scheme and Coding Method for Multi-view Video Coding

Multi-view video allows us to experience more realistic feelings with arbitrary view-points and a wide range of scenes, because those are captured by multiple cameras simultaneously but at different positions. Since the amount of data to be processed increases in proportion to the number of cameras, we need to develop efficient coding methods. In this paper, we propose an efficient view interpolation scheme and coding method for multi-view video using the correlation between views. After generating an intermediate view image, we use it at the encoder as an additional reference frame using the proposed ´VIP P-picture´ coding method. The proposed view interpolation method improves the generated inter-view image about 1~4 dB and the ´VIP P-picture´ coding method improves coding efficiency about 0.5 dB.