Optimized techniques for video compression and enhancement

• Main Author: Nguyen, Viet Anh
• Other Authors: Tan Yap Peng
• Format: Theses and Dissertations
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: https://hdl.handle.net/10356/19311
• Institution: Nanyang Technological University

Motivated by the requirement of efficient techniques to process the increasing amount of today’s video data for better and faster visual communication services, the research work in this thesis is mainly concerned with the optimized techniques for video compression and enhancement. On the one hand, we explore the solutions to optimize the video encoder and transcoder to reduce the complexity. Specifically, a fast block-matching algorithm for motion estimation is proposed based on the integral frame attribute concept for video encoder, while efficient methods for transcoding video between the existing compression standard and the latest H.264/AVC standard are proposed for video transcoder. On the other hand, we also consider the quality degradation issue affected by these coding and transcoding solutions in order to enhance the decoded video by post-processing. We address a new research problem by blindly enhancing the quality of the video reconstructed from multiple compressed copies of the same video content. By using quantization constraint information derived from the multiple compressed copies and resembling each AC coefficient distribution by Laplacian/Cauchy models, we can reconstruct a video that achieves better quality than any of the available copy.