Multi-Objective Optimization for Image Segmentation Based on the Tsallis and Rényi Entropies
Salah Eddine Mechkouri, Saleh El Joumani, Lhoussaine Masmoudi · International journal of imaging and robotics · 2019
Multi-objective optimization is one of the techniques increasingly implemented for image segmentation. In this paper, we present a new thresholding technique based on the Tsallis and Renyi entropies for the satellite image segmentation. This technique is part of the multi-objective optimization approach. We will compare this technique with the one previously developed by El Joumani et al [4]. This technique allows the use of the multi-objective optimization approach to find the optimal thresholds regarding two criteria: Tsallis criterion and Renyi criterion. The evaluation of our method is built on the Levine and Nazif criterion and the mean squared error. It has been tested on synthetic images, and applied to the segmentation of a very high spatial resolution satellite image of the desert environment. The selected zone of study is located in Laâyoune-Sakia-el-Hamra province, in the South-West of Morocco. Our approach has given a satisfactory result.