Efficient Technique for Segmentation of Various Objective Functions using Bio Inspired-HPSO Algorithm

Abstract

Image segmentation is very challenging due to complex work. Researchers created a number of bio-inspired algorithms to calculate the optimal threshold values for segmenting such pictures. Extending them to multilayer thresholding, their exhaustive nature makes them computationally costly. Using the Particle Swarm Optimization (PSO) technique, the author presents a computationally efficient image segmentation approach, termed hybrid PSO (HPSO). Improved segmentation quality was achieved by using the HPSO method, which required more computing time. MSE, ME, PSNR, Entropy, CPU time, FSIM, CPU timing, and SSIM were measured for all instances studied. For image segmentation, the suggested HPSO method has been shown to be the most promising and computationally efficient approach to date. The proposed work has been implemented in MATLAB simulation tool. The results obtained through simulation indicates better accuracy compared to the existing baseline approaches. In addition, the suggested method surpasses others in achieving stable global optimal thresholds, according to a study of its convergence rate. Image processing, remote sensing and computer vision applications may benefit from the findings of this investigation.