Quantization Parameter Selection Method for H.264-based Multi-view Video Coding

In this paper, we propose a QP (Quantization Parameter) selection method for H.264 based MVC (Multi-view Video Coding). The proposed method adaptively calculates QP for B frames in anchor (BANC frames). This method utilizes rate-distortion costs of the two frames which are referred by BANC frames. Using these costs, we estimate the correlation between the reference frames. The first reference frame is coded as I or P frame, and then the second reference frame is coded as P frame referring to the first reference frame. We assume that the rate-distortion cost of the second reference frame is smaller as the two reference frames are more highly correlated. QP for BANC frames is then calculated based on the correlation. Experimental results show that the proposed method improves coding efficiency of the reference prediction structure presented by Fraunhofer-HHI for MPEG core experiments. The proposed algorithm achieves 0.07~0.16 dB of average PSNR gain.