A study on Differential Evolution and Cellular Differential Evolution for multilevel color image segmentation

Image segmentation is the process of partitioning an image into multiple regions and is widely utilized in various domains to ease separating and searching particular objects within a given image. Multilevel thresholding is one of the most important techniques used for segmenting images. However, selection of optimum thresholds is still a challenging problem. Since evolutionary algorithms are being applied in literature over the decade for improving the accuracy and computational efficiency of multilevel thresholding methods, in many cases, such stochastic algorithms may prematurely converge into local optima and lack accuracy and stability. Structured Evolutionary Algorithms are proposed to overcome these limitations. Therefore the main aim of this research work is to propose a multilevel thresholding method based on Cellular Differential Evolution for color images segmentation. We present in this paper a comparative performance study to evaluate the efficiency and the accuracy of Standard DE and Cellular DE. We also present the results of this comparative study which shows that the structuring of the population within DE improves its performance in terms of accuracy, robustness and speed convergence. The proposed algorithm is also evaluated by comparing it with three other algorithms using a well-known benchmark images. Such comparison reflects the efficiency of our algorithm.