Vision perceptual based rate control algorithm for multi-view video coding

Multi-view video coding technique is widely applied on 3D technology. The view number is increased to improve the quality of 3D viewing experience. However, the data storage requirement is very huge for the view number incensement. The rate control is very important issue to trade off the quality and bit-rate. This paper proposes a rate control method which utilizes Human Visual System (HVS) to distribute the bit-rate in the difference vision perceptual region. The interesting area for 3D viewing experience is determined to arrange appropriate quality and bit-rate on multi-view video coding. In the same bit-rate coded video bitstream, the interesting region in the reconstructed image by the proposed algorithm has better image quality then by the JMVC.

I. Introduction

Three-dimension (3D) video has become a major topic in television system [1], providing the viewer offering a new level of reality in their viewing experience. Multi-view video coding (MVC) [2] is an essential aspect of 3D video applications, which include multiple video sequences captured by several cameras at the same time from different viewpoints [3] resulting in enormous quantities of coding data. However, the high transmission bandwidth, the huge data storage requirements, and the time needed for the complex computation involved increase the difficulty involved in making 3D video a reality for home TV users. An efficient compression algorithm for 3D video is crucial. MVC has been considered in several video coding standards [4]–[6], including MPEG-2 multi-view profiles, MPEG-4 multiple auxiliary components and H.264. Recently, Joint Video Team (JVT) developed the joint multi-view video coding (JMVC) [7] to serve as a common platform for research into MVC.