Improved logarithmic search for efficient video compression

Abstract

For efficient video compression, the BMAs should be designed and implemented in such a way that that the Motion Estimation takes very less time and will have lesser complexity. In this paper, two new algorithms namely Improved Logarithmic Search (ILS) and Zero motion predicted ILS (ZMILS) are proposed which are based on the basic idea of Three Steps Search (TSS) algorithm. In these BMA, the search pattern of proposed algorithms changes the efficiency both in terms of complications required for each macro block (MB) and quality of the compensated images. We improved the TSS and NTSS by changing the searching pattern of locations in reference frame and by adding no movement early prediction case with these algorithms we further reduced the computations. The computations required by the new developed algorithm are lesser per macro block and even 50% of few existing techniques while retaining the quality of the reconstructed image.