Multilevel image segmentation using hybrid Darwinian swarm Optimization

DPSO-FOHA I and DPSO-FOHA II algorithms, based on multilevel thresholding are proposed in this paper. Optimal multilevel thresholds for colored images are maximized by using Otsu's between class variance functions. The Darwinian principle has been used to improve the value of fitness function along with the concept of fractional calculus, which optimizes it in lesser number of search iterations. Comparative analysis is presented between existing and proposed algorithms for performance assessment using standard deviation, PSNR, SSIM and computational search time of CPU. Experimental results depict that DPSO-FOHA II outperforms DPSO-FOHA I and another state of art methods in terms of PSNR, SSIM, fitness function and computational time.