Journal of Multimedia Processing and Technologies Peiying Li, Zhongtang Huo 17 1 2026 https://doi.org/10.6025/jmpt/2026/17/1/11-20 https://www.dline.info/jmpt/fulltext/v17n1/jmptv17n1_2.pdf This paper proposes an improved image segmentation model that combines the K-means clustering algorithm with Particle Swarm Optimization (PSO) to enhance computer vision performance. Traditional K-means suffers from sensitivity to initial cluster centers and high computational complexity, especially in RGB color space. To address these issues, the authors integrate PSO to perform a global search for optimal initial cluster centers before applying K-means for local refinement. This hybrid approach avoids local optima, reduces computational load, and accelerates processing speed. Experiments compare the proposed model against standard K-means and PSO-K models using three color images. Results show the enhanced model achieves the shortest segmentation runtime across all test images and delivers superior edge detail and segmentation accuracy. The study highlights that while image complexity affects processing time, the proposed method maintains consistently high efficiency and precision. The authors conclude that their model significantly improves image segmentation in computer vision tasks, making it more suitable for real world applications in fields like agriculture, healthcare, and smart systems. By leveraging color space conversion and intelligent optimization, the model demonstrates robustness, faster convergence, and better handling of fine image details compared to existing techniques.