الفهرس 
H.264 VIDEO CODING STANDARDOnly 14 pages are availabe for public view
 |  | from | 200 |  |  |
| | | from | 200 | | |
Abstract
In real time applications such as video streaming and video conferencing it is important that the video encoding/decoding is fast. H.264/MPEG-4 Part 10 or AVC (advanced video coding) is currently one of the most widely used industry standards for video compression. It has been developed to achieve significant improvements, in the compression performance, over the existing standards.
In fact, the high compression performance is accomplished by the implementation of intra and inter prediction techniques that remove spatial and temporal redundancies. Most video coding standards depend heavily on the use of inter frame prediction technique to obtain high compression efficiency. Inter prediction technique is developed by motion estimation (ME) and motion compensation (MC) techniques.
It is known, that most of the complexity of the H.264/AVC compression standard lies in the H.264 encoder. Motion estimation (ME) introduces high computational complexity. It consumes about 70% of the time required for H.264 encoder.
Several studies have been provided to reduce the complexity of the H.264 encoder and the compression time without affecting the quality of the video. In other words, many ME algorithms has been designed and implemented to overcome the problem of complexity and time consumption of the inter prediction technique.
In this thesis, a MATLAB design and implementation of Enhanced Predictive Zonal Search algorithm (EPZS) ME algorithm for H.264 AVC is presented.
This thesis also introduces a comparative study on the most popular ME algorithms using MATLAB. It presents that the EPZS algorithm is superior on the most common ME algorithms. from studying the ME algorithms utilized in the JM reference software “18.6”, EPZS ME provides the best results in terms of speed, video quality and data rate but there is no consideration about how it can be effectively implemented in hardware.
This thesis proposes, a novel hardware design and implementation of EPZS ME algorithm for H.264 AVC is presented. An efficient implementation of inter prediction process using EPZS algorithm as the motion estimation technique is realized. The resulting motion vectors are compared with that of the exhaustive search ME algorithm to verify its efficiency. The proposed design of EPZS is implemented in “VHDL”, simulated using “ModelSim SE 6.5”, synthesized using “Xilinx ISE Design Suite 13.3” and verified using “SIMULINK”. The proposed architecture provides a high speed and low hardware complexity implementation of H.264 encoder permitting its use in real-time applications of H.264 /AVC standard.
Also this thesis introduces two enhancement methods on the EPZS algorithm. These enhancements provide an improvement on the speed of the EPZS performance by about 18.05% without affecting the quality of the output frame.