Multi-objective optimization for worldview image segmentation funded on the entropies of Tsallis and Rényi

Autor: Lhoussaine Masmoudi, Saleh El Joumani, R Zennouhi, Salah Eddine Mechkouri
Rok vydání: 2020
Předmět:
Zdroj: Multimedia Tools and Applications. 79:30637-30652
ISSN: 1573-7721
1380-7501
DOI: 10.1007/s11042-020-09572-4
Popis: Image analysis usually refers to processing of images with the objective of finding objects presented in the image. The extraction and the analysis of image data is a fundamental step for image segmentation, in this work a new method allowing the evolution of the threshold satellite image defined and based on the optimization multi-objective for segmentation of Worldview images and funded on the Tsallis and the Renyi entropies. The Objective is the reclassification of all unclassified pixels by the previous method in 2017. The improved analysis and the optimized method multi-objective thresholding are proposed. First, respectively for each Worldview image selected, the optimal thresholds for all the criteria used in this study is find. Finally, by using the evaluation criteria corresponding to the Levine and Nazif criteria and the criteria of the mean square error, in order to challenge the performance of this method to that previously developed in 2017. The results obtained by this approach were very satisfactory and the efficacy of this method confirmed. This method overcomes the difficulties of the method previously developed in 2017 and obtained results that are more precise. Therefore, the new method based on multi-objective optimization contribute significantly to performance.
Databáze: OpenAIRE
Externí odkaz: