Based On The Theory Of Graph Partition Image Segmentation Technology Research

Image segmentation is one of the basic techniques in the field of image processing and computer vision, image segmentation refers to the use of the image grayscale, color, texture, shape, an image is divided into several independent objects have a common attribute of pixelsthe essence of a clustering process in accordance with the pixel property. Image segmentation based on graph partitioning theory is the image map into a weighted directed graph, and then use the mature theory of graph theory to image segmentation. This paper studies the image segmentation based on graph partitioning theory, the specific work done mainly in the following areas:(1) This article outlines the background and application of image segmentation and image segmentation of the mathematical theory based on the introduction and description. Described in detail several typical image segmentation method, and the advantages and disadvantages of these typical image segmentation method are analyzed and compared. Focus on image segmentation method based on graph partitioning theory. The method is to map the node set of the map image, the image pixels corresponding to the weighted graph, the relationship between the corresponding pixels between the adjacent edge set, the attributes of the node corresponding pixel feature information (such as grayscale, color, etc.), the side of the property corresponding to the difference or similarity between the pixels.(2) The minimum spanning tree-based image segmentation algorithm. First discussed in detail the minimum spanning tree algorithm; On this basis, the minimum spanning tree algorithm is applied to which the image segmentation, image segmentation method based on the minimum spanning tree, minimum spanning tree image segmentation algorithm efficiency and the advantages and disadvantages of carried out a detailed analysis on this basis, the minimum spanning tree image segmentation algorithm has been optimized, and experimental results show that the optimized algorithm can get better segmentation results.(3) Based on weeks algorithm for image segmentation. This is a fast segmentation method based on graph partitioning theory can be applied to the segmentation of still images and moving target, set a reasonable basis points, to be completed by the automatic segmentation. Peer weeks algorithms on the basis of a detailed overview of an original isoperimetric algorithm to improve the algorithm implementation steps, and finally gives the corresponding experimental results and experimental analysis of experimental results improved isoperimetric algorithm is a fast, stable, good segmentation method based on graph partitioning theory. (4) Based on normalized cut image segmentation method.This is an unsupervised image segmentation techniques, it does not require initialization, both the similarity between the different groups within the same group similarity to maximize a global criterion. This paper summarizes the advantages and disadvantages of the normalized cut algorithm, and on this basis, an optimal normalized cut algorithm, experimental results show that the normalized cut algorithm optimized to be more satisfied with the image segmentation results.Finally, a summary of this paper and the future research direction of the image segmentation method based on graph partitioning theory outlook.