Performance evaluation and improvement of H.264/AVC video over UMTs.

This thesis evaluates the performance of H.264/AVC video transmitted over the third generation mobile network. The aim of this research is to enhance the robustness of H.264/AVC video in wireless applications. The thesis is started with a brief review of video coding tools, including Discrete Cosine Transform (DCT), quantisation, Inter prediction coding and entropy coding. Then we focus on the main features of the newest standard H.264, also called "Advanced Video Coding" (AVC). The concept of levels and profiles with its applications are defined. The fundamental differences and similarities between this codec and the previous H.263 series are highlighted. Special interest is also paid to the syntax of video bit stream and the error resilience techniques in H.264/AVC. In Chapter 4, the complexity of H.264 coder is assessed in terms of processing time consumption. The effect of encoder parameter settings, including the entropy coding and rate control algorithm is evaluated. Then, H.264 reference codec is enhanced in order to decode the video stream delivered through highly error prone environments. Furthermore, the quality of H.264 video stream transmitted over Universal Mobile Telecommunication Systems (UMTS) upon different encoder parameter sets is studied, and optimum parameter configurations for test sequences are derived. Two novel techniques are proposed to enhance the robustness of video performance under error prone environments. They both are based on Intra refresh techniques to stop error propagation, one at Group of Blocks (GOB) level and the other at MacroBlock (MB) level. In the first method at GOB level, two GOBs in neighbouring frames are chosen to be Intra refreshed. In the second method, two criteria are combined together to select Intra MBs, namely the Sum of Absolute Difference (SAD) and the size of each MB. The achieved improvement in the proposed methods is shown on both the objective and subjective scales. Finally, conclusions are drawn in Chapter 6. Suggestions for future work are also given