Digital Image Segmentation Using Delaunay Triangulation Algorithm

Abstract

Image segmentation is the process of partitioning a digital image into multiple segments according to some homogeneity criterion. Different approaches are suited to different types of images and the quality of output of a particular algorithm is difficult to measure quantitatively due to the fact that there may be much “correct” segmentation for a single image. Although the field encounters several challenges, this research work seeks to present a segmentation algorithm for gray intensity images. Image features extraction is first performed to obtain the approximate, if not exact, representation of a given image. Image enhancement was also carried out by exploiting the principle of histogram equalization which is one of the well-known image enhancement techniques. The proposed approach is based on the Delaunay (DT), one of the techniques in computational geometry, which generates clusters of intensity values using information from the vertices of the external boundary of triangulation. This makes it possible to produce segmented image regions. The algorithm was tested on sampled images before and after enhancement and the production compared. Our algorithm successfully extracts cogent features from a given image and made use of the Delaunay Triangulation to carry out segmentation. The method is robust, precise and independent of translation, rotation and scaling and that makes the result quite good for image segmentation. Image segmentation based on Delaunay triangulation technique seem to be a promising future research area with respect to its applicability to underwater mine and submarine detection, better target recognition, medical imaging as well as to automated computer vision.