Research On Active Contour Model Based Simultaneous Segmentation And Registration Method

Segmentation and registration are key issues in image processing, and they are widely used in many application fields, so both of them have received extensive concern. But because of the independence of segmentation and registration researching, and neglected their relation-ship, traditional segmentation and registration methods can't get satisfying processing result. Along with the extending field of application and the improving of image processing precision demand, this problem has come to be more serious.Active Contours has become an important method of image processing. In this paper, Along with the correlative technique——variational method and level see method, we present a active contour model based Simultaneous segmentation and registration. The key idea is to build an energy function that combined the active contour based image segmentation to the contour feature based image registration by introducing the registration mapping into active contour model. And then though minimizing the energy function using a gradient flow, we can achieve the goal of simultaneous segmentation and registration. On the one hard, this new method can reduce the influence of noise on the original images and lead to improved registration, on the other hand it also can improve the precision segmentation, especially for segmentation the blurred images. The results of the experiments are satisfied, and indicate the validity the validity of the method.Non-rigid image registration is a difficult problem of image analysis. In order to make this method be applicable to non-rigid registration, based on the same joint segmentation and registration idea, an improved simultaneous segmentation and registration method is presented in the back of the paper. By estimating non-rigid registration parameter field between two images and using the level set method, the improved method not only achieved the goal of simultaneous segmentation and registration, but also improved the ability to deal with the topological changes. Experimental results show that the improved method is well performed.