Research For Fitting Energy Segmentation Technology Based On Level Set

With the development of digital technology, medical imaging technology has become one of the most important means in clinical medical. In order to effectively make use of medical image information, medical image segmentation technology has been gradually entered into the field of medicine. As an important part of medical image processing, medical image segmentation has a certain practical value in respect to separating and extracting the tumor targets or tissues.Recently, people has proposed thousands of segmentation methods with specific theoretics. Since Level set method was proposed as an mathematical theory for effectively solving the evolutionary questions of curves and surface, it had gradually becoming a research focus in the field of image segmentation, and is extensively applied in other processes of Image Processing.The thesis studies the image segmentation technology based on Level set, realizes the segmentation algorithms of gray and color images based on two models, and solves the problems that can't segment images with inconsistent intensity. The main content of this thesis is as follows:Firstly, introduces the mathematical principles of Level set method, analyzes on the merits and demerits of the traditional method, discusses the problem of reinitialize level set function.Secondly, discusses the Mumford-Shah model and its simplified model. Through studying deeplyChan-Vese model based on Level set, I realize the segmentation of grey images based on the model, and presents the color images segmentation algorithm. In accordance with the experimental results, I conclude the merits of the algorithm, analyses the reasons that this model can't segment images with inconsistent intensity, and proposes the solutions.Thirdly, introduces the Local expanded fitting energy model based on Level set, presents the segmentation algorithm of gray and color images based on this model, and solves the drawback ofChan-Vese model. The experimental results shows, this algorithm could segment precisely images with inconsistent intensity, and segment betterly the medical images with complicated structure.