Research On Image Segmentation Algorithm Based On Variational Method

The variational method has become a hot issue in the theory of image segmentation is due to the modeling process is relatively easy, the extension of the methods is better, the process of achieving is simple. At the present, the variational method has become one of effective and widely used methods in the image segmentation processing. In order to improve the accuracy and robustness of segmentation results, and reduce the time complexity of image segmentation, the variational level set method is used to implement image segmentation done a more in-depth discussion. The use of variational level set method for image segmentation processing has been discussed in detail, and two kinds of variational method are proposed to achieve image segmentation.Because the existence of the traditional active contour has not very good effect on image segmentation robustness, in this thesis, an active contour model combined region-based model with gradient-based model are proposed using the image entropy energy and gradient sum. Firstly, image entropy and image gradient are used as weighting factors, and the segmentation models are constructed by use of the contour models based on image gradient and the contour model based on image region.Secondly, the level set is embed in the above models, so that the segmentation process can be continuous and also adapt to the topological changes. Finally, in order to improve the speed of the curve evolution, the narrow band method is used to make the activity curve evolve rapidly.The experimental results show that the proposed model have a better robustness with faster evolution speed.The time complexity of traditional active contour segmentation is sometimes not very good. In this paper, an active contour model is designed based on contour neighborhood gradient. The model can be used to fit the iteration function by calculating the neighborhood gradient value of the contour line, and then the time complexity of the segmented image is reduced. The experimental results can demonstrate that the proposed algorithm can segment images quickly, and have a smaller time complexity.Two kinds of modified variational techniques can improve the quality of image segmentation, and provide technical support for the application of image segmentation technology in biological engineering, medical field, industrial design and so on.