Fuzzy Clustering Based Image Segmentation Algorithms

Fuzzy Clustering Based Image Segmentation Algorithms

M. Ameer Ali (East West University, Bangladesh)
DOI: 10.4018/978-1-60566-908-3.ch013
OnDemand PDF Download:
No Current Special Offers


Image segmentation especially fuzzy based image segmentation techniques are widely used due to effective segmentation performance. For this reason, a huge number of algorithms are proposed in the literature. This chapter presents a survey report of different types of classical and shape based fuzzy clustering algorithms which are available in the literature.
Chapter Preview

1. Introduction

The application of digital images is rapidly expanding due to the ever-increasing demand of computer, Internet and multimedia technologies in all aspect of human lives, which makes digital image processing a most important research area. Digital image processing encompasses a wide and varied field of applications from medical science to document processing and generally refers to the manipulation and analysis of pictorial information. Image processing is mainly divided into six distinct classes: i) Representation and modelling, ii) Enhancement, iii) Restoration, iv) Analysis, v) Reconstruction, and vi) Compression. Image analysis embraces feature extraction, segmentation and object classification (Baxes, 1994; Duda & Hart, 1973; Gonzalez & Woods, 2002; Jahne, 1997; Jain, 1989), with segmentation for instance, being applied to separate desired objects in an image so that measurements can subsequently be made upon them.

Segmentation is particularly important as it is often the pre-processing step in many image processing algorithms. In general, image segmentation refers to the practice of separating mutually exclusive homogeneous regions (objects) of interest in an image. The objects are partitioned into a number of non-intersecting regions in such a way that each region is homogeneous and the union of two adjacent regions is always non-homogeneous. Most natural objects are non-homogeneous however, and the definition of what exactly constitutes an object depends very much on the application and the user, which contradicts the above generic image segmentation definition (Gonzalez & Woods, 2002; Karmakar, Dooley, & Rahman, 2001; Spirkovska, 1993; Haralick & Shapiro, 1985; Fu & Mui, 1981).

Segmentation has been used in a wide range of applications, with some of the most popular being, though not limited to: automatic car assembling in robotic vision, airport identification from aerial photographs, security systems, object-based image identification and retrieval, object recognition, second generation image coding, criminal investigation, computer graphic, pattern recognition, and diverse applications in medical science such as cancerous cell detection, segmentation of brain images, skin treatment, intrathoracic airway trees, and abnormality detection of heart ventricles (Karmakar, Dooley, & Rahman, 2001; Pham & Prince, 1999; Liu, et al, 1997; Pal & Pal, 1993).

Different applications require different types of digital image. The most commonly used images are light intensity (LI), range (depth) image (RI), computerized tomography (CT), thermal and magnetic resonance images (MRI). The research published to date on image segmentation is highly dependent on the image type, its dimensions and application domain and so for this reason, there is no single generalized technique that is suitable for all images (Pal & Pal, 1993; Karmakar, 2002).

Complete Chapter List

Search this Book: