Edge Detection and Medical Image Segmentation Using Metaheuristics

Abstract

In the past few years, image processing has emerged as an active area of research in different domains including medical imaging, satellite imaging, computer vision etc. In any computer vision algorithm, edge detection and segmentation are certainly the most significant tasks that require high accuracy detection. In the field of medical imaging, computer-aided segmentation is in high demand as the manual segmentation is subjective and prone to errors. Thus, the present work is devoted to these two most perennial tasks namely edge detection and medical image segmentation. However, designing accurate, reliable and practically usable techniques for edge detection and medical image segmentation is a real challenge to deal with, owing to the presence of noise and high variability in medical images. newlineIn spite of considerable progress in this domain, extracting edges from images corrupted with high-density impulse noise and performing segmentation on medical images still lie as open research problems. Therefore, in this thesis, these problems are addressed and handled using metaheuristic algorithms along with the image processing techniques. Metaheuristics algorithms have the capability to deal with the challenges associated with edge detection and medical image segmentation which are otherwise difficult to be solved by the traditional approaches. These problems are summarized as three main research objectives which are covered in five chapters. newlineThe first objective deals with the detection of edge map in images with low and high-density impulse noise using ant colony optimization (ACO) and bird swarm algorithm (BSA) respectively. The techniques are implemented on Berkeley Segmentation Dataset (BSD) and standard image dataset. Smooth and continuous edges are obtained under different noise variations as compared to existing techniques. newlineIn the second objective, segmentation of two important regions of retinal fundus image namely optic cup and optic disc is performed.