A Comparative Study on Various Image Segmentation Techniques

International Journal of Research and Scientific Innovation (IJRSI) | Volume VI, Issue VII, July 2019 | ISSN 2321–2705

A Comparative Study on Various Image Segmentation Techniques

 C. M. Naga Sudha¹, T. Brinda²

¹Teaching Fellow, Anna University, MIT Campus, Chennai- 600 025, India
²AP/CSE, Anna University Regional Campus, Tirunelveli-627007, India

Abstract: – Medical images(CT scans, MRI scans) play a vital role in diagnosing the disease. In order to diagnose the disease, medical images must be viewed clearly. So for clear visual appearance, medical images have been splitted into Region of Interest(RoI) and Non-Region of Interest(RoI) in which RoI must be visualized clearly for diagnosing the disease and to provide preventive measures. Therefore, to focus only on the RoI which is diagnostically important, segmentation is used. Segmentation is the most fundamental and important technique which is used to analyze images. The main aim of segmentation method is to simplify the complex problem into simpler ones by dividing the pixels based on the characteristics. After segmenting the image into regions, the pixel information can be stored and transmitted without any loss. Hence, there is a need for developing segmentation algorithms which consumes minimal time. In this work, existing segmentation methods such as edge-based segmentation, region-based segmentation, threshold-based segmentation, clustering-based segmentation have been analyzed based on the performance.

Index Terms: Segmentation, Medical image, Edge-based segmentation, Region-based segmentation, threshold-based segmentation, clustering-based segmentation

I. INTRODUCTION

Computer Vision is an interdisciplinary field in which the automation of the task can be performed by extracting, analyzing and understanding useful information from an image or sequence of images. Image segmentation is a technique in which the images are segmented and processed individually. It is the process of partitioning the digital images into multiple segments which can be used to locate objects and boundaries. In other words, we can say that segmentation can also be used to extract the foreground from the background. Image segmentation is the prerequisite process before analyzing the images. In some cases, image de noising is performed before segmentation which is a challenging task. Selecting the most appropriate method on image segmentation is necessary to perform any task. Assigning pixel values to the correct segment is very important and it remains a tough task also. The number of segments to be made in the image depends on the size of the whole image which is to be segmented. Classifiers can be used inside the segmentation method which can be used to divide and analyze the images based on their structures. Segmentation has its own applications in measuring the size and shape of the image. Content-based image retrieval is one among them deals with the image segmentation.

Read More