Multi-objective Image Segmentation Based on Gradient Vector Flow Field Analysis

In this paper, the generating mechanism of GVF field and the characteristic of GVF in real contour position are analyzed. Based on that, the sufficient conditions of zero field are defined and initial contours are gotten automatically which objects are interested by the users and the distance about 3-5 pixels from real contour. The suggested method solves the problem of multi-objective image segmentation in GVF snake model, improves the accuracy of extraction, insensitive to noise, reduces the time complexity and allows the users to choose the interesting objects autonomously.