Adaptive Fast Search Block Motion Estimation In Video Compression

Abstract

With the advancement of telecommunication technologies, such as internet, video conferencing and HDTV, we need an effective video compression technique. Motion estimation and motion compensation are the most complicated and time consuming part of any video coding technique. Motion estimation helps to reduce temporal redundancy that exists between successive video frames. The motion estimation part of any video codec should be such that, it can reduce computational complexity without having any effect on the quality of the video. The motion estimation process can be more efficient if we use spatial and temporal correlation between the blocks in a frame and between two consecutive frames. In this thesis, a new search method for block motion estimation in video has been presented that uses neighbouring blocks of current macro-block and the block in the previous frame having the same coordinates as that of current macro-block for prediction of motion vectors. In the proposed method we use the motion vectors of neighbouring blocks that are more likely to be helpful in the prediction process. By using these motion vectors a search centre is located, around which a search window is placed. In this thesis, we have introduced Sorted Search Method (SSM) algorithm for motion estimation and compared the performance with existing Block Based Motion Estimation (BBME) techniques. Different sorting search methods have been developed by taking different neighbours around the current macro - block and their performances are compared.