The Research And Application Of Image Segmentation Method Based On Level Set Theory

Image segmentation is one of the most fundamental and important task in image analysis, It aims to divide an image into several coessential regions, so as to extract the interested object in the image. Image segmentation is the precondition of most image analysis and image understanding tasks. Experienced decades of development, various theories and methods have been applied to image segmentation method, among them, active contour method has been paid more and more attention in recent years. Compared with other segmentation method, active contour method can provide continuous with sub-pixel precision segmentation results. In addition, the application of level set theory, makes the realization of active contour method more flexibility and convenience.Active contour method can be mainly divided into two classes:edge-based method and region-based method, In essence, these two kinds of methods make use of the local information and global information respectively. However, we found in the experiment that, simple utilization of one particular type of information is insufficient to guide the evolution of segmentation contour correctly. Therefore, in this paper, we try to combine the local information and global information of image reasonably, in the context of active contour method, it is combining local fitting energy and global fitting energy dynamically in the energy functional of the active contour, so as to guide the evolution of segmentation contour more reasonably.In this paper, we firstly introduced the role image segmentation plays in the image project, and briefly introduced the history of the development of image segmentation and some kinds of typical segmentation method; then, we mainly introduced the theoretical basis of active contour method and level set method, and the representative level set based active contour methods. Finally, after the analysis of the role that local information and global information plays in the image segmentation, we gave our segmentation energy model, and analyzed the meaning of each part in the model, and we also gave the method of how to realize the model.For several test images, we compared the results of our new method and the results of some other active contour methods, through the analysis of experimental results, the rationality and validity of our method is verified.