Research Of Motion Estimation Algorithm For Video Compression Applications

High-definition television (HDTV) and net video are widely used with the rapid development of modern digital communication technology. Digital video compression is the key technology for mass video transmission with limited network bandwith, and is becoming the wordwide academic focus.To follow the word research step, this thesis study on the video compression based on motion estimation. The main contents are as follows:1.Firstly, We introduces the basic technologies in video compression, such as the macro-block size,prediction of initial search point,search strategy and so on. Then, we researche and analyze the classics of the block-matching motion estimation algorithm. The performances of the classic algorithms, such as motion estimation time, bit rates and PSNR-Y, are test with three videos and the advantages and disadvantages of the motion estimation algorithm are explained.2. According to the defect of easy to fall into local optimal, the UMHexagons and diamond search algorithms are improved, tests show that the improved algorithm have a better motion estimation efficienty and PSNR. Moreover, an adaptive fast search algorithm is proposed based on the intensity degree motion. Performances are test compared with NTSS, FSS, BBGDS and DS, results shows that addaptive Fast search algorithm can improve the speed of search algorithm markedly without loss of aeeuraey.