Motion estimation for video compression and decompression techniques.

: Ala'a Basil Hussein
Dr. Sarcot N. Abdullah
Dr. Oday A. Ridha

Abstract: The digital video coding is a rapidly evolving field. Although its applications are numerous (from TV broadcasting to video conference), bit rate reduction is always needed .This has led to intensive efforts toward effective video compression techniques. In this work an adaptive codec scheme of H.263 standard that depends on transform coding (DCT) and quantization process in compression is proposed. In addition motion estimation search algorithms were used to estimate the predictive frames. The traditional search algorithms (Three Step Search, Cross Search and Two Dimensional Logarithmic Search) were applied. An adaptive mechanism was proposed (which is a hybrid of TSS 2DLog and Cross search) and implemented to handle the time delay associated with all searching methods. However the minimizing in time of search is about (38% less from TSS, 5% less from 2DLog, 1.5% less from Cross search). The compression ratio using the suggested technique (ranges from 26% up to 64%) of the original size video can be achieved according to condition of compression process. Testing some video samples (uncompressed AVI format) representing several cases changing scene shows the effect of increasing and decreasing reference frames (I-frame) on compressed video quality (represented by PSNR), estimation time and compression ratio. Also the effect of changing video quality was tested to consider its effect on compression ratio, time of estimation and PSNR. The developed proposed simulation programs are implemented using MatLab 7.6.0 (R2008a) computer language and executed for testing purpose on computer with processor Pentium 4 (1.6 GHz dual core), 1G byte of RAM under Windows XP sp2 operating systems.