A fusion of remote sensing images segmentation based on Markov random fields and fuzzy c-means models
Remote sensing images segmentation is a challenging task in analysis process of terrestrial applications. In this paper, we propose a combination of two segmentation methods of remote sensing images. The first based on MRF (Markov Random Fields) method which takes into account the neighboring labels of the pixels and the second is computed with a Fuzzy C-means technique to improve the likelihood criterion. After, a fusion by Dempster Shafer theory is performed on results from the two images segmentation techniques. The contribution of this work is to improve the belongingness of pixels in order to extract more useful information in terrestrial applications of remote sensing images. The whole algorithm is evaluated on a real remote sensing image and experimental results show that the developed approach has more performance than previously discussed methods in term of accuracy and quality of segmentation.