Graph Theory Based Approach For Image Segmentation Using Wavelet Transform

This paper presents the image segmentation approach based on graph theory and threshold. Amongst the various segmentation approaches, the graph theoretic approaches in image segmentation make the formulation of the problem more flexible and the computation more resourceful. The problem is modeled in terms of partitioning a graph into several sub-graphs; such that each of them represents a meaningful region in the image. The segmentation problem is then solved in a spatially discrete space by the well-organized tools from graph theory. After the literature review, the problem is formulated regarding graph representation of image and threshold function. The boundaries between the regions are determined as per the segmentation criteria and the segmented regions are labeled with random colors. In presented approach, the image is preprocessed by discrete wavelet transform and coherence filter before graph segmentation. The experiments are carried out on a number of natural images taken from Berkeley Image Database as well as synthetic images from online resources. The experiments are performed by using the wavelets of Haar, DB2, DB4, DB6 and DB8. The results are evaluated and compared by using the performance evaluation parameters like execution time, Performance Ratio, Peak Signal to Noise Ratio, Precision and Recall and obtained results are encouraging.