Multi-objective optimization for worldview image segmentation funded on the entropies of havrda–charvat and exponential

Saleh El Joumani,Khalid El Amraoui,Salah Eddine Mechkouri,Rachid Zennouhi,Lhoussaine Masmoudi
DOI: https://doi.org/10.1007/s11042-024-18988-1
IF: 2.577
2024-03-28
Multimedia Tools and Applications
Abstract:This work presents a new thresholding technique supported by the Havrda criterion and exponential entropy for Worldview image segmentation to advance satellite image classification. This technique, part of a multi-objective optimization approach, makes it possible to find optimal thresholds as a function of criteria such as the HAVRDA-CHARVAT criterion and the exponential entropy criterion. To demonstrate the importance of this study, we compare these results with those obtained by the method developed by El Joumani et al. in 2017 in our research laboratory. Certainly, the experimentation of the new algorithm on a sample of selected images, representing the desert environment of southwest Morocco, showed good results. The Levine and Nazif criterion, the Rosenberger criterion, and the root-mean-square error criterion are available for evaluating this method.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering
What problem does this paper attempt to address?