Edge Based Region Growing

Abstract

Image segmentation is a decomposition of scene into its components. It is a key step in image analysis. Edge, point, line, boundary, texture and region detection are the various forms of image segmentation. Two of the main image segmentation techniques edge detection and region growing are highly in use for image segmentation. In human visual systems, edges are more sensitive than other picture elements. Edge detection technique when used alone for image segmentation results in small gaps in edge boundaries. It is sensitive to local variations intensity and the contours obtained are usually not closed. Region growing technique when used alone results in errors in region boundaries and the edge pixels might be joined to any of the neighboring pixels. Edge based region growing corresponds to the optimum image segmentation technique in which the both edge detection approach and region growing approach is integrated. This technique is based on the fact that edge based and region based approaches are complementary to each other and use ancillary information to guide the segmentation procedure. This segmentation procedure separates the image in two segments namely background and foreground. The algorithm described here is for integrating edges and regions. Firstly, the edge map of image is obtained by using canny edge operator. Then the edge region is grown. Very small regions are removed by merging. Thus the effect of noise is completely eliminated. The two types of seeds (pixels) hot and cold are obtained in the edge region and according to the type of data being analyzed and application area, the image is segmented into background and foreground objects. It offers very precise segmentation in detecting objects of different sizes and also non-rigid targets. This approach is not sensitive to the parameters, such as the sizes of different operators and thresholds in the edge detection and edge region detection. The algorithm is implemented in MATLAB and the result demonstrates that the algorithm is robust, satisfying and work well for images with non-uniform illumination.