Image Segmentation Using Kapur’s Entropy Maximization

Image segmentation plays a vital role in Medical Image Processing and Computer Vision. This process of partitioning a digital image into multiple segments and assigning a label to every pixel of the image is definitely a challenging task. Several segmentation methods have been proposed in the literature, of which thresholding techniques remain the most widely adopted choice. In thresholding, a set of proper threshold values is selected to optimize a functional criterion. One such functional criterion is based on Kapur's Entropy. As previously known, evaluation complexity of multilevel thresholding remains superior to bi-level thresholding. The proposed work attempts to achieve the same by employing Electromagnetism-Like optimization approach that exhibits interesting search capabilities while maintaining low number of function evaluations. The results of the algorithm are compared against Cuckoo Search Optimization. Experimental analysis demonstrates that the proposed approach improves segmentation.