Color image segmentation by fuzzy morphological transformation of the 3D color histogram

Summary form only given. We present a color image segmentation method using fuzzy mathematical morphology operators on the 3D color histogram. Segmentation consists in detecting the different modes which are present in the 3D color histogram and associated to homogeneous regions. In order to detect these modes, we show how a color image can be considered as a fuzzy set of pixels characterised by its membership function to a mode. It is defined by the concavity analyses of the histogram. Then, a fuzzy morphological transformation is applied to this membership function in order to enhance the modes. The efficiency of our proposed fuzzy morphological approach is then illustrated using a color image.