Textured images segmentation by a multiresolution morphological decomposition method

This contribution deals with the textured images segmentation. The model exploits morphological operators and order filters properties. A morphological decomposition filters bank is built to isolate elementary patterns by decomposing the textural image characteristics. The 1 and 2 order statistic moments and the gradient means are computed in order to select the best feature component image which allows to perform the image segmentation. The method is illustrated by a real image randomly textured.