Perceptual Awareness Rate Control for Multi-View Video Encoder in Stereoscopic Display

Abstract

Multi-view video coding technique has been widely applied to 3D stereoscopic display. This paper proposes a rate control algorithm for multi-view video codec which is a tradeoff method between the data compression ratio and picture quality of 3D video. This study utilizes the characteristics of visual perception for 3D video viewers to determine the interesting regions in all view videos. Interesting regions with different levels, which depend on the existence of macroblocks with high motion, objects with individual shape, or obvious texture, are determined in the proposed algorithm by considering the corresponding Just Noticeable Distortion function, luminance histogram bin variation and edge strength, respectively. Then the proposed algorithm assigns adequate quantization parameters to control the bit-rate for different levels of interesting regions such that the video quality of the interesting regions is preserved and the bit-rate in the non-interesting regions is reallocated to 3D video storage and transmission for stereoscopic display. The experiment results show that the proposed method successfully achieves high quality preservation and low bit-rate for 3D effects.