Based On The Improved GVF Research Image Segmentation Problem

Traditional Snake model tries to position the boundary of targeted contour,in order to ensure the continuity and smoothness of the contour curve,which adopted the up-down information processing method instead of the traditional bottom-up information transportation.It is more convenient and efficient to use Snake model to extract certain interested target,which has been widely used since the day it was invented.However,there lay some flaws in this model,such as small capture scope and Sag area boundary extraction,which in some extent have been improved by GVF method formulated by Xu etc.GVF method also has its limits,such as imperfect extraction of deep sag area boundary.To solve this problem, this paper puts forward VGVF method as an improvement.The work of this paper and the conclusion is as follows:(1) GVF is an improved method based on the traditional Snake model,which gives detailed explanation on the gradient vector flow principle and implementation method of the mathematical formula.Compared with the traditional Snake model,the experiments indicate a great improvement on setting of the initial position and sag area,which in some extent improved the extraction of sag area boundary.(2) A new method of improvement is put forward.On basis of GVF Snake field,this method contributes to the GVF vector product composite field with a constant vector cross product.If the constant vector is a unit vector,we will generate a new force field it is called rotation gradient vector field(Namely VGVF Snake method).The most significant feature of the VGVF Snake method is that the direction of its power goes around the edge of targets.(3) The dynamic programming algorithm and greedy algorithm have been analyzed in the paper,which contribute to the calculation of the energy function minimization. Combined the VGVF method with the dynamic programming algorithm and greedy algorithm, the experiment results have been analyzed. Compared with the experimental results of GVF method,it has been clarified that the improvement can solve the extraction problem of deep sag area image edge.The experimental results show that the improvement can be conducted successfully.