Multi-mode weight selection method for joint prediction based coding

This paper presents an efficient multi-mode weight selection method for joint prediction based coding. This scheme makes use of multi-mode joint prediction of motion estimation/disparity estimation (ME/DE) instead of existing three-mode prediction for reconstructing a target image. Moreover, a fast ME/DE approach is introduced in order to improve the coding speed based upon the correlation between motion vectors and disparity vectors. The major features of the proposed approach is that not only its computational load is decreased drastically for calculating disparity vectors, but also it can realize more flexible and adaptive coding than existing interpolated ME/DE algorithms. Experimental evaluation is performed using benchmark and self-generated stereo image sequences. The simulation results show that the proposed scheme can achieve better image quality than existing similar algorithms.