| Multimedia applications have become an integral part of our daily life through their use in mobile communications, video conferencing, entertainment and more. Multimedia includes video and audio data transfer, of which video data communication is a more complex process and consumes greater bandwidth. Hence, developing low complexity/low-power methods to transmit, receive and store video data has always been an area of immense research. The state of the art video coding technique is called the H.264 coding standard, proposed by the Joint Video Team [1]. Several methods have been proposed to optimize the complexity of the H.264 coding process with a graceful degradation in video quality. This thesis describes a method for reducing the complexity of the H.264 coding process while maintaining compliance to the existing standard and yielding an acceptable video quality.;The proposed method introduces a pre and post processing block to the H.264 black box which enables identification and classification of important information in a given video frame in the time frequency domain. The entire H.264 coding process is performed only on the highly informative portion of a given video frame, while the motion information extracted from them is used to code the rest of the frame. This enables the user to bypass some of the time consuming and complex operations while coding the less informative content of the frame. This time-frequency domain based approach facilitates the use of spatial domain coding techniques thereby ensuring compliance to the standard. The proposed method is compared to the H.264 standard in terms of PSNR, speed up, bandwidth, area and power. The proposed method can be suitably tuned for both high and low bit rate applications for varying resolutions of the video frame. |
