Active Contour Model And Its Application And Research In Medical Endoscope Image Segmentation

The image information has become an important way of getting the information, and it is extremely important to deal with the image information and make use of it. Image segmentation is one of the key issues in the field of the image processing and the computer vision, and it is also the important component of the image analysis and the visual system. A variety of image segmentation methods have been proposed. The complexity of the medical image makes the medical segmentation very bad. The first problem of the computer-aided medical diagnostic system to be solved is image segmentation.Because of their own limitation, traditional image segmentation methods may get incomplete and inaccurate boundaries of the partition. The image segmentation based on active contour model,i.e. Snake model, can combine the bottom level visual attributes of the image with a priori knowledge of the partition target, and it can obtain a complete expression of the divided region. But active contour model is mostly based on image borders, and its stability is difficult to meet the requirements of medical image segmentation; and traditional Snake model can not change its topology easily, and it is not adapt to the variability of the lesion number. So to get a better segmentation result, active contour model still needs to be improved.The main work that this thesis has done is as follows:Analyze the principles and methods of the parameter deformation model, and through experiments compare the external forces of various models in the aspect of the performance, the advantages and disadvantages, and the impact of the initial outline to the convergence result; give the theory of geometric active contour model, and analyze and test the advantages and disadvantages that the CV method, and make a small improvement to the CV method; propose a new application of the image segmentation, that is video stream segmentation, and achieve it; analyze the CV method for color image segmentation, and make the experiment and give the results.