Image Segmentation Using Snake Model with Quadtree

摘要

This paper presents a news approach to image segmentation using Snakes model in conjunction with Quadtree. The proposed method firstly employs Quadtree to divide images into small blocks. Regions with image detail will be segmented into blocks with smaller size, and the background of the image will be assigned larger block size. After the preprocessing of Quadtree segmentation, those salient features in the image such as edges, lines and region boundaries will be reserved as a part of the result. The rest of the regions without significant features will be further processed using principal components analysis to find a vector such that the projected values from vectors may maximally preserve the variances among vectors. Then, Snakes have been used extensively in locating object boundaries. The Experiments illustrate the proposed method has yield good results where objects can welll be extracted from the background.