Block Matching Algorithms for Motion Estimation: A Performance-Based Study

Motion estimation (ME) process is the most crucial and time-consuming part of video compression technique. So many block-based motion estimation techniques were developed to make ME easy and fast. In this paper we have reviewed almost all the existing BMA from very old Full Search (FS) to recently developed Reduced Three-Step Logarithmic Search (RTSLS) and Cross Three-Step Logarithmic Search (CTSLS), and so on. We have also compared them based on the computations needed per macroblock and the PSNR value of compensated image. Earlier Adaptive Rood Pattern Search (ARPS) was found to be most computationally efficient but during the review we have applied the old algorithms along with recently developed Zero Motion preadjusted RTSLS (ZMRTSLS) and Zero Motion preadjusted CTSLS (ZMCTSLS) are found to be more computationally efficient than even ARPS.