Research Of Watershed Algorithm And Level Set Algorithm On Medical Image Segmentation Application

Medical image segmentation is a very active branch in the field of imagesegmentation study, its purpose is split interesting image area that is isolated from thesurrounding environment by extracting the characteristics of the target.Medical image segmentation results are directly related to the Good or bad effect ofclinical diagnosis and the accuracy of the treatment; the split speed is related to theability to meet real-time clinical needs and provide real-time interactive tools forclinicians; the robustness of the algorithm are directly related to the ability to maintainits segmentation accuracy and reliability in complex clinical environment; degree ofautomation is directly related to the efficiency of the algorithm in actual clinicalapplications, and is also a reliable guarantee of accuracy.In this paper, through the research of level set algorithm and watershed algorithmbased on the advantages and disadvantages of the traditional level set method andwatershed algorithm in medical image processing, making algorithms improved. Thinkingout a fast level set algorithm and improved watershed algorithm to achieve better resultsin medical image segmentation.First of all, through the study of the watershed algorithm, for the serious problem ofover-segmentation and sensitivity to noise shortcomings of watershed algorithm inmedical image segmentation, presenting an improved watershed algorithm to achievebetter results, Effective solving a serious problem and noise-sensitive shortcomings inmedical image segmentation solution.Second, the analysis and comparison of several existing level set algorithm model toidentify a level set algorithm LBF model that having operating efficiency andsegmentation results in general the best method.by changing the kernel function of LBFmodel, to greatly accelerate the running speed of the LBF model and to get a rapidimprovement LBF model.Finally, for segmentation results LBF model being effected by the characteristics ofthe initial evolution curve, the paper extract the maximum connected region from thewatershed segmentation results as a initial evolution curve of the LBF model,then,combining the watershed algorithm and level set algorithm to segment a largenumber of medical images. Experiments verify the feasibility of the algorithm, andthrough the analysis of its operating efficiency and segmentation results, the algorithm isan effective method that solutes the problem of the choice of the initial evolution curve,fixes segmentation results，at the same time，achieve good effect of segmentation.Meanwhile, since no manual selection of the initial evolution curve, so the degree ofautomation of the algorithm is significantly improved.