图像分割的图论方法综述

A SURVEY ON GRAPH THEORY APPROACHES OF IMAGE SEGMENTATION

图像分割是图像处理与计算机视觉领域的基本问题之一,其本身固有的不适定性是该领域研究的最大挑战。图像分割的图论方法充分利用图像的整体和局部特性,具有很大的灵活性,较高的计算效率及良好的分割特性,成为分割领域的一个新的研究热点。根据当前主要的几类基于图论的分割模型概括了图像分割图论方法的基本框架,包括图的映射和构造、分割准则及目标函数的设计及求解。系统综述了图像分割图论方法的每一类别的理论及研究进展。最后就图像分割图论方法中尚存的问题及未来的可能发展方向提出了见解。

Image segmentation is one of the fundamental problems in image processing and computer vision area. Its inherent ill-posedness is the greatest challenge of this research field. The graph theory approaches of image segmentation make full use of. global and local properties of the image, have quite big flexibility, higher computation efficiency and better segmentation property, and become the new focus of research in segmentation area. According to several major graph theory-based segmentation models at present, in this paper we summarise the basic framework of the graph theory approach of image segmentation, including the image mapping and construction, segmentation criteria as well as the design of and the solution of the target function. We give a systematic survey about the theories and research progress of each category in regard to graph theory approach of image segmentation. Finally, we offer our view on the problems existing in graph theory approach of image segmentation and the possible development direction in the future.