A Performant Clustering Approach Based on An Improved Sine Cosine Algorithm

Image segmentation is a fundamental and important step in many computer vision applications. One of the most widely used image techniques clustering. It process segmenting intensities non-homogeneous into homogeneous regions based on their similarity property. However, clustering methods require prior initialization random centers often converge to local optimum, thanks choices initial centers, which major drawback. Therefore, overcome this problem, we improved version sine-cosine algorithm optimize traditional improve results. The proposed method provides better exploration search space compared original SCA only focuses best solution generate new solution. ISCA able speed up convergence avoid falling optima by introducing two mechanisms that take account first given position second found so far balance exploitation. performance approach was evaluated comparing several algorithms metaheuristics such as SCA, genetic (GA) particle swarm optimization (PSO). Our evaluation results were analyzed fitness values metrics paper, demonstrates high gives satisfactory other comparison methods.