An enhanced hough transform method in image segmentation for circle detection

Abstract

The rapid growths of the computer technologies of the world-wide web have increased the amount and the complexity of multimedia information. Images are the most widely used media to represent various shapes. Image segmentation is the essential step to localized shapes in an image. Image segmentation can be further categorized into three components namely; image thresholding, edge-based segmentation and region-based segmentation. Hough transform is one of the edgebased segmentation technique used to detect object of circular shape. However it was found that Hough transform is unable to detect a circular shape with unknown radius and it is also unable to detect semi-circular shape. Thus, Hough transform is enhanced to solve these problems by examining Hough peaks in the Hough matrix. Different shapes display unique distribution and pattern of Hough peaks. The results reveal that Hough peaks for semi-circular shape appear on the same x and y coordinate, and Hough peaks for circular shape appear on the different x and y coordinate. Thus, it is feasible to utilize this method to locate a circular, semi-circular or other shapes appear in the image like single line, square object, ellipse and rectangular.