Computation-aware algorithm based on inter-view motion vector analysis for multi-view video coding

To provide higher compression efficiency with a computational constraint, this paper proposes a computation-aware algorithm based on inter-view motion vector analysis for multi-view video coding. First, the correlation of motion vector differences and rate distortion costs between different views are analyzed. Based on the analytical results, the motion vector difference and rate distortion cost of the macroblock in the main view are considered to allocate the computation resources under the constraint of computational availability. Furthermore, a global disparity vector (GDV) normalization algorithm is applied to smooth the global disparity vector for the purpose of shifting the information of the main view to the auxiliary view so that the shifted result can be as accurate as possible. Experimental results prove that our proposed algorithm can maintain a quality performance and restrain the bit-rate increasing with low and limited computation resource support.