Color Image Segmentation Using Fuzzy Clustering and Fusion: Application to Biomedical Images

Abstract

In this paper, a new color image segmentation method based on Fuzzy clustering and data fusion techniques is resented. The proposed segmentation consists in combining the three component images (R, G and B) of the same image, to gather, in order to increase the information quality and to get an optimal segmented image. In the first step, the segmented images are obtained by applying fuzzy c-means clustering to the information's coming from the three independent component images. In the second step, on the obtained segmented images with specific primitive colors, a combination rule is used to integrate the segmentation results over the three-color components. Experimental investigations and comparative studies with the other previous methods are carried out showing thus the robustness and superiority of the proposed method in terms of medical and textured image segmentation. Index Terms: Segmentation, biomedical image, fuzzy c-means, fuzzy fusion, conflict, and data fusion. I.INTRODUCTION he image segmentation is the process of partitioning an image into homogeneous regions. The goal of segmentation is to locate objects and boundaries in images [1] [2]. More precisely, image segmentation is the process of assigning a label to every pixel in an image such that pixels with the same label share certain characteristics, such as intensity, color, tone or texture, etc. However, processing images in levels is easy and fast compared to color images. Nevertheless, the color image processing [3] is a very important task when the objects cannot be extracted using gray scale information but can be extracted using color information. The color images processing is one of the important areas. It used in different domains and many techniques have been proposed. Color images can be represented using several color models such as RGB, HSI, YCbCr etc. [3]. It is also possible to convert between color models using the linear and non-linear transformations of the RGB color space. Each color representation has its advantages and disadvantages [4] [5]. There is still no color representation that can dominate the others for all kinds of color images yet. The major problem of color image segmentation is the high correlation and spatial redundancy of multi-band image histograms and the difficulty of clustering using multi-dimensional histograms. Hence, the general segmentation problem consists in choosing the adopted color model in a specific domain [5] [6]. In the past, many authors have addressed the color image segmentation problems using different methods [7] [13] have proposed a segmentation method based on fuzzy sets and Dempster-Shafer (DS) evidence theory. The idea is to assign, at each image pixel level, a mass function that corresponds to a membership function in fuzzy logic. The membership degree of each pixel is determined by applying the FCM algorithm to the gray levels of the image. Then, the DS combination rule and decision are applied to obtain the final segmentation. However, the major problem in the use of Demp...