A new fuzzy segmentation approach based on S-FCM type 2 using LBP-GCO features Lotfi Tlig n , Mounir Sayadi, Farhat Fnaiech SICISI Unit, ESSTT, 5 Av. Taha Hussein, 1008 Tunis, Tunisia article info Article history: Received 14 December 2010 Accepted 4 March 2012 Available online 12 March 2012 Keywords: Texture segmentation Gabor filter Local binary pattern Fuzzy clustering abstract Gabor filtering is a widely applied approach for texture analysis. This technique shows a strong dependence on certain number of parameters. Unfortunately, each variation of values of any parameter may affect the texture characterization performance. More- over, Gabor filters are unable to extract micro-texture features which also have a negative effect on the clustering task. This paper, deals with a new descriptor which avoids the drawbacks mentioned above. The novel texture descriptor combines grating cell operator outputs derived from a designed Gabor filters bank, and local binary pattern features. For the clustering task, an extended version of fuzzy type 2 clustering algorithm is also proposed. The effectiveness of the proposed segmentation approach on a variety of synthetic and textured images is highlighted. Several experimental results on a set of textured images show the superiority of the proposed approach in terms of segmentation accuracy with respect to quantitative and qualitative comparisons. Crown Copyright & 2012 Published by Elsevier B.V. All rights reserved. 1. Introduction Most natural surfaces exhibit textures. In the case of an image, it defines the spatial relationship between the gray-scale values of the pixels in a region of the image. Texture segmentation plays an important role in pattern recognition and computer vision. Many fields of applica- tion are concerned with it, including robotics, remote sensing, medical imaging, etc. [1]. Textured regions are usually characterized by two-dimensional variations in intensity, which makes the mathematical modeling of textures for the segmentation process very difficult [2]. The segmentation task is to divide a given image into a set of disjoint regions with homogeneous parameters, usually denoted as features, based on a clustering algorithm. It can be defined as an identification of different regions with uniform textures, or as an identification of the boundaries between them. In image analysis literature, there are already a large number of supervised and unsupervised segmentation algorithms. The only difference between both approaches is that the supervised one needs a prior knowledge of textures present in the image, while the unsupervised one is implemented without using knowledge texture sam- ples. There is no optimal technique that could handle all the segmentation of different types of image. Since Bezdek [3] proposed the fuzzy c-means (FCM) clustering algorithm, his technique has been applied in a wide variety of image processing applications [4–6]. Let us start with a brief overview; conventional FCM is an unsu- pervised algorithm, which has been proven effective in many image segmentation applications. However, this clustering technique has got two disadvantages. The first does not include any spatial information in the image. The second, FCM is considered as type 1 fuzzy clustering approach which employs a membership function (MF) graduated by crisp numbers in the range [0,1]. The use of an accurate MF to make decision for something uncertain is not reasonable. In recent years, to reduce the negative effects of uncertainty, type 2 fuzzy logic sets have become more and more applied Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/image Signal Processing: Image Communication 0923-5965/$ - see front matter Crown Copyright & 2012 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.image.2012.03.001 n Corresponding author. Tel.: þ216 96661099. E-mail address: tliglotfi@yahoo.fr (L. Tlig). Signal Processing: Image Communication 27 (2012) 694–708