Image segmentation algorithm based on cumulative residual Tsallis entropy

Abstract

An algorithm which applies Cumulative Residual Tsallis Entropy (CRTE) to threshold image segmentation is proposed. Cumulative residual Tsallis entropy is an uncertainty measure generalized from Shannon entropy. The approach utilizes the cumulative residual Tsallis entropy function as the objective function, specifies the threshold segmentation criterion, and finds the optimal cumulative residual Tsallis entropy function solution. This method is applied to Berkeley Segmentation Dataset and compared with several classical segmentation methods on the same dataset. Experiment findings suggest that this algorithm is capable of producing improved segmentation results.