The Research Of One New GVF-Snake Model

This paper research and improve Gradient Vector Flow Active Contour model . This paper introduces the development and several traditional image segmentation methods of Active Contour Model: image segmentation method based on threshold, image segmentation method based on edge detection, image segmentation method based on region, image segmentation method based on the clustering analysis, segmentation method based on texture of image. This paper also simply introduced basic principle of Parameter -Active Contour model and Geometrical Active Contour Model. It also analyzes the active contour model is better than the traditional method of excellent properties. Image segmentation, image contour detection is an important part of image processing, the medical image contour detection will be able to test from visual into automatic monitoring, this is lightened the doctor's work pressure. This will indirectly boosted the development of medical technology.This paper analysis the advantages and disadvantages of traditional Gradient Vector Flow Active Contour model.A new concept:normal vector-gradient vector angle,was put forward.By using the new concept,gradient vector diagram was amended. Euler equation , finite difference method and the law of cosines is important mathematical tools of the model. The basic principle of the effective gradient vector flow is that curve evolution of normal direction is also an influence factor of external force field which can made the curve evolution more accurate. Adaptive balloon force was quoted in order to speed convergence. Processing of the new model is more cautious than the original one, and propertyof anti-noise was improved. This new method needs to calculate a group of cosine value of new normal vector-gradient vector Angle each iteration, and judge whether the compatibility degree Angle arrive threshold. This paper uses law of cosines deduced cosine value of normal vector-gradient vector Angle , and mades a detailed mathematical derivation.Comparative experiments by using the new model , Active Contour model, and Gradient Vector Flow Active Contour model ,has been done,which handling U-shaped figure, medical heart internal profile MRI diagram, chest transverse MRI images. Experiments show that the new model has its own advantage compared to traditional GVF Active Contour model: the new model can detect medical image's contour which is more closer to the real contours, ability of anti-noise raises.