Bit allocation techniques for video coding and broadcasting applications

• Main Author: Yao, Wei
• Other Authors: Susanto Rahardja
• Format: Theses and Dissertations
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/54879
• Institution: Nanyang Technological University

In today's world, it is no longer uncommon to see video applications that employ a variety of transmission and storage systems, thanks to the significant progress in video coding technologies together with great improvements on network infrastructure and the increasing growth of storage capacity and computing power. With an increasing demand for digital video applications, the expectation on the quality of these applications is getting higher. Bit allocation is a classical problem in video coding and it plays an essential role in achieving better coding efficiency and better visual quality. Usually bit allocation can be considered as the constrained optimization problem, in which a given bit rate budget is distributed among different units in different application scenarios. In this thesis, the bit allocation problem for various video applications is studied. The focus of the research is on algorithm design and coding structure optimization. The goal is to provide an efficient way to utilize the available resources so as to produce optimum visual quality of the reconstructed video with the given resource constraint. Bit allocation within a single video source is more commonly referred to as rate control. If there is a limited memory buffer at the encoder, rate control of an encoded video will reduce buffer overflow. Rate control also ensures that the output bit rate of video stream remains within the capacity of transmission channel. When conventional rate control algorithms perform bit allocation at group of pictures (GOP) level, a fixed GOP size is usually adopted without considering the correlation between frames. A new rate control algorithm in which the sizes of GOPs are adapted to the video content is proposed. Temporal redundancies among frames can be better removed and it transforms to improvements on coding efficiency. Scalable video coding (SVC) provides a scalable scheme for "one-source-multiple-use" in heterogeneous network. The layered structure of SVC allows one single bit stream to serve multiple users with different decoding and display capabilities and/or network channels with different bandwidths. Both cross layer bit allocation for individual user and bit allocation for multiple users are studied in this work. A combined SNR and spatial scalability scheme is proposed to optimize a specific layer according to customer composition. A motion refinement scheme is then developed to further improve the coding efficiency of that specific layer through a better bit allocation between motion and residual information. When bit allocation is unbalanced in current coding structure, an improved inter-layer prediction scheme is proposed for layers with both spatial and quality scalability and a new concept of auxiliary layer is introduced to further improve the coding efficiency. In broadcast applications, multiple video sources are encoded in parallel and transmitted simultaneously over a bandwidth limited channel. Coding of each video independently at an equal proportion of the channel bandwidth is a straightforward method to allocate channel bandwidth but it suffers from big quality fluctuations between videos as well as within each video. A new joint bit rate allocation scheme is designed to dynamically allocate the channel bandwidth so that a more uniform video quality is achieved for various videos and the quality variation within each video is also minimized. With recent advances in display and camera technologies, 3D video are becoming popular for next generation broadcast and streaming applications. Although there are various representation formats available for 3D video, stereoscopic seems to be the most mature technology at this moment. However, due to the increased video data compared to conventional single-view video, efficient coding format and joint bit rate allocation among multiple stereo videos on bandwidth limited transmission links are still challenging problems. A scalable coding scheme for stereo video is first designed such that regulating output bit rate of each video to match the allocated bandwidth can be achieved by simple bitstream extraction. The bit allocation problem for stereo videos is more complex as it includes both bit allocation between videos as well as between views. By exploiting human visual system, a joint bit allocation scheme is developed to do adaptive bit allocation between two views so that higher perceptual quality in each video is achieved while the variance of perceptual quality among different stereo videos is also reduced comparing with existing method.