The heuristic algorithm based segmentation procedures are widely used to find optimal thresholds for RGB and Gray scale images. In this paper, Otsu based bi-level and multi-level image segmentation is carried for a class of gray scaled images using Improved Particle Swarm Optimization (IPSO). Optimal thresholds for the test image are attained by maximizing Otsu’s between-class variance function. The performance of the proposed IPSO based segmentation procedure is validated with the existing methods, such as Particle Swarm Optimization (PSO) and Bacterial Foraging Optimization (BFO) algorithms. The performance assessment between algorithms are verified using well known image parameters such as objective function, Peak to Signal Ratio (PSNR), and the Structural Similarity Index Matrix (SSIM). The result shows that for most of the images, IPSO based method offers enhanced result compared to the alternatives.