A New Objective Function Based Multi-Level Image Segmentation Using Differential Evolution

This Paper represents a multi-level image thresholding approach based on the normalized index value of image and probability of pixel intensities. One new objective function proposed, which is the multiplication of normalized index value and probability, to obtain the scenario. This multiplication measure is then optimized to obtain the thresholds of the image. In order to solve an optimization problem, Differential Evolution (DE) as a meta-heuristic approach is used, which results a fast and accurate convergence towards the optimal solution. The performance of DE is compared to other well-known optimized algorithms like Particle swarm optimization (PSO), Genetic Algorithms (GA). The outcomes of images are compared with Kapur entropy, Tsalli entropy and Otsu method, both visually and statistically for establishing the perceptible difference in image.