Small Population Based Modified Parallel Particle Swarm Optimization for Motion Estimation

Abstract

In this paper, the authors propose a small population based modified parallel particle swarm optimization (SPMPPSO) and its application to reduce computational time for motion estimation in video sequence. In motion estimation, initial search, search space, matching criteria, search parameter and step size are important aspect to predict the position of the current macro block for which motion vector is to be found. In the proposed technique, the position equation of PPSO known as step size is modified to find best matching block in current frame. In the SPMPPSO, small population i.e. five swarms is used to find global optimum value. Due to neighbourhood search criteria (N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> ), the convergence is very fast. The limitations of existing methods like computational time, search parameter, initial search and search space are overcome by SPMPPSO. The suggested method saves computational time up to 94% when compared with other published method. The SPMPPSO can be used in adaptive network, self-managing system ubiquitous learning environment etc for efficiency improvement.