Segmentation of color images using a linguistic 2-tuples model

In this paper we address the problem of color image segmentation transforming it into a decision making paradigm. A set of experts is provided, so that each expert assigns a preference degree of each pixel to every object of the image considering also the ignorance associated with such assignation. We represent the objects by means of fuzzy linguistic labels and using the decision-making model based on 2-tuples we apply an aggregation phase to classify each pixel. To obtain the segmented image we consider the preference values associated with a pixel and also with its neighbors. We test our segmentation method on Berkeley segmentation database [21] and we compare the experimental results with the ones obtained by FCM method [4] and MAP-ML method [17].