A hybrid motion compensation for wide and fixed field-of-view image sequences

In this paper, a rotation-aware motion compensation method for wide and fixed field-of-view(FOV) image sequences is presented. Most predictive models of conventional video compression standards such as translational motion models perform well for narrow FOV cameras like television cameras but not for wide FOV cameras. Therefore the image quality of rotationally moving scenes are significantly degraded after the compression process with bandwidth limitation. The proposed rotation-aware predictive model is designed for wide and fixed FOV cameras of mobile phones, and the main objective of the proposed method is to reduce the variance of motion vectors. Experimental results show that the proposed method reduces the variance of motion vectors while the image quality is maintained or improved.