Study On Image Segmentation Based On Snake Model

Image segmentation is a key technology from image processing to image analysis and it is the basis of image understanding. Image segmentation based on Snake model is a highly efficient image segmentation technology. It can effectively utilize local information and global information of the image, accurately locate boundary and maintain linear smooth. Because of these advantages and its profound theoretical background, image segmentation based on Snake model became a hotspot in computer vision field as soon as it is proposed.In this thesis, a review of image segmentation is presented. It includes the background, the purpose, significance and development of image segmentation. And then, the Snake model including its backgrounds, theories, algorithm, advantages and disadvantages is discussed.Aimed at the weakness that Snake model cannot converge concavity district correctly, an improved Snake model is proposed by modifying the continuity energy and the area energy in Snake model. In algorithm realization, the two-step greedy algorithm is taken in. Experiment results indicate that the improved Snake model can be better converged into the concavity district of image.Referring to some disadvantages mentioned above, a new force is put forward and is defined as inner product force, and then a new kind of energy function with inner product force is designed. The inner product force makes the Snake model constringe to the deep concavity district and the more complex district rapidly without judging the concavity and convexity property of the outline points. As the inner product force can produce a contractile energy, which can cross local noise points, it can overcome the shortcoming that the traditional Snake model can easily constringe to local noise points. Simultaneously, the new energy can also reduce the segmentation result's dependence on initial point.