Abstract:

To alleviate 3D-HEVC computational burden, a novel depth level range determination method based on motion homogeneity is proposed in this paper. Since the multi-view videos show the same frame at the same time, it is important to use the selected depth levels and prediction modes to save coding time since they are highly multi-view content-dependent. The motion homogeneity model for CU is characterized by motion vectors and prediction modes with corresponding CUs in the reference views. According to this model, we propose a fast depth level range determination method. The experimental results show that the fast method is able to reduce the computation time by an average of 30.9%, while maintaining a similar rate-distortion (RD) value compared to the initial 3D-HEVC system.