An Efficient Image Segmentation Approach Based on Graph Theory

Image segmentation technology refers to a basic operation for image processing, and it can provide preparation works for high-level image analysis. The goal of this paper is to segment images with high accuracy and efficiency using the graph theory. In this paper, we propose an image segmentation algorithm based on graph cuts. We convert the image segmentation problem to a labeling problem, and we aim to allocate each pixel or block a label by deal with a graph optimization problem. The main idea of this paper lies in that we introduce some external information in the graph cut based image segmentation. Firstly, we create an augmented image which integrates the original image with texture features. Secondly, we propose a novel method to combine the region and boundary information in our proposed graph cut based image segmentation algorithm. Experimental results prove that our proposed algorithm can achieve lower average error rate than other methods, especially for images which contain salient objects and simple backgrounds.