Visual Quality-Regulated Three-Dimensional Video Coding (3-DVC)

Abstract

This chapter reviews (in Sect. 3.1) issues associated with research and development of three-dimensional (3-D) video coding in the context of the current push towards better viewing quality and experience of visual communications, broadcasting, and entertainment represented by ultra-high definition (UHD) television (TV), three-dimensional digital video (3-DV), multi-view video (MVV), and free viewpoint TV (FTV). It highlights (in Sect. 3.2) an agonizing impasse in a much-needed paradigm shift for video coding design from a bitrate-driven to a visual quality-driven design approach [23, 24, 64] based on Shannon’s entropy [44] and rate-distortion (R-D) theories [4, 45], making true the viewing quality and experience promised by the aforementioned visual communication technologies. The key is to deliver with precision a designated visual picture quality discernible by human viewers for an intended application, maximizing intended visual experience at no more cost than theoretically and practically necessary by design rather than by chance or “the best effort.” It examines (in Sect. 3.3) the state of the art in three key areas closely related to perceptual 3-D video coding, including 3-D video coding (3-DVC), perceptual video coding, and visual quality assessment and perceptual quality/distortion metric design. Issues associated with design of visual quality-regulated 3-DVC are discussed (in Sect. 3.4) for the transition from bitrate-driven design to visual quality-driven design, identifying a number of challenges for further research and investigations. It provides (in Sect. 3.5) insights into a comprehensive mathematical formulation and illustration of a theoretical framework for perceptual quality-regulated encoding of 3-D video in terms of just noticeable differences (JNDs) and techniques applicable to various coding components of the framework.