Particle Swarm Optimization on Parallel Computers for Improving the Performance of a Gait Recognition System

Abstract

In recent years, the gait recognition (GR) using particle swarm optimization (PSO) algorithm (OSO) has been execute very fast and accurate with single computer, but with the appearance of parallel computing (PC), it was necessary to use this technique to improve the results of GR. This study presents the use of parallel computing approaches (PCA) to implement PSO for a GR system (GRS) to decrease processing while maintaining reconstructed image quality. These approaches are: Codistributor and parallel cluster. Many experiments have been executed with recognition between the two approaches. The experimental results showed that increasing the PSO swarm size, decreasing number of iterations, and increasing number of workers used for the PCA can reduce recognition time and increase performance. Best results were obtained from implementing parallel computing with eight workers and 100 iterations. The execution time reached 4s and PSNR reached 44db. At the same time, the best results were obtained from PCL approach.