Performance Evaluation of Selected Chaos-Enhanced Particle Swarm Optimisation for Image Segmentation

Abstract

Image segmentation remains a fundamental challenge in computer vision, where multilevel thresholding becomes computationally demanding as the number of thresholds increases, and traditional optimisation methods often converge prematurely. This study integrates chaotic maps into the Particle Swarm Optimisation (PSO) algorithm to enhance population diversity, prevent premature convergence and improve segmentation quality. The aim is to evaluate three chaos-enhanced PSO variants (Logistic+PSO, Tent+PSO and Skew Map+PSO) for image segmentation using multilevel thresholding on the Berkeley Segmentation Dataset (BSDS500). Chaotic sequences were used to replace the pseudo-random number generators employed for population initialisation, parameter adaptation and position updates, and the three chaotic maps were integrated into the PSO framework. The models were implemented in Python and evaluated on the preprocessed BSDS500 dataset using Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Feature Similarity Index (FSIM), convergence speed and success rate.The Skew Map+PSO achieved the best results with a PSNR of 26.78 dB, SSIM of 0.9212, FSIM of 0.9089, success rate of 97.2% and convergence in 36.8 iterations, followed by Tent+PSO (25.67 dB, 0.9056, 0.8923, 95.8% and 42.3 iterations) and Logistic+PSO (24.89 dB, 0.8892, 0.8745, 94.2% and 48.5 iterations).The study establishes that chaos enhancement significantly improves PSO performance, with Skew Map+PSO recommended for high-accuracy applications such as medical imaging, Tent+PSO for balanced performance and Logistic+PSO for real-time systems. The chaos-enhanced PSO models can therefore serve as reliable, robust solutions for multilevel-thresholding image segmentation.