Multi-view video coding based on resultant vector estimation

Aiming at fully exploiting the temporal redundancy and spatial redundancy, we propose a multi-view video coding scheme based on resultant vector estimation. Combining the disparity vector and motion vector as one vector, the novel "resultant vector" was first proposed. Meanwhile, by extending the loop constraint to multi-view images for a parallel camera geometry model, redundancy between views is greatly reduced by utilizing the linear relationship of adjacent views and the "resultant vector". Experimental results show that our scheme\´s performance is superior to conventional methods such as GOP (group-of-picture) or M-picture. Coding efficiency is improved about 1.2~1.5 dB.