A multilevel thresholding method based on HPSO for the segmentation of various objective functions

Abstract

Image segmentation is very challenging due to its complex working. Researchers created a number of bio-inspired algorithms to calculate the optimal threshold values for segmenting such pictures. Extending them to multilevel thresholding, their exhaustive nature makes them computationally costly. Using the Particle Swarm Optimization (PSO) technique, the author presents computationally efficient image segmentation approach, termed hybrid PSO (HPSO). Improved segmentation quality was achieved by using the HPSO method with less computing cost & time. MSE, ME, PSNR, Entropy, CPU time, FSIM and SSIM were measured for all instances studied. For image segmentation, the suggested HPSO method has shown to be the most promising and computationally efficient approach to date. 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.