Otsu Multi-Threshold Image Segmentation Based on Improved Geese Swarm Algorithm
Threshold segmentation is a widely used image segmentation technique.However,the tradi-tional Otsu algorithm based on maximum interclass variance faces challenges in multi-threshold image seg-mentation such as high computational complexity,long execution time,and insufficient segmentation ac-curacy.To address these issues,a multi-threshold image segmentation algorithm based on improved geese swarm optimization algorithm(CBLSGSO)is proposed.This algorithm embeds the Cubic Chaos Mapping model into the initialization process of the gravitational search algorithm to enhance population diversity.A multi-region guided structure is introduced to dynamically divide the population and design different ev-olutionary mechanisms,expanding the population optimization range.Adaptive cosine strategy and butter-fly algorithm search strategy are incorporated to improve the algorithm's convergence accuracy,effectively balancing global and local optimization capabilities.To verify the performance of the improved Otsu algo-rithm,ACC,Jaccard,Specificity,F1-score,FSIM,SSIM,and PSNR are selected as evaluation metrics and compared with image segmentation algorithms proposed by different scholars in recent years to vali-date the effectiveness of the algorithm.The experimental results demonstrate that the Otsu image segmen-tation method based on the improved algorithm can more quickly and accurately solve complex image seg-mentation problems.
万方数据
维普
