Active Contour Model Integrating Global and Local Information

Abstract

Active contour model based on global information cannot segment intensity inhomogeneity images effectively. Although active contour model based on local fitting can segment images with intensity inhomogeneity effectively , they are senstive to the initial contour and often fall into local minima. Based on the above issues, we proposed active contour model integrating global and local information in this paper. First of all, we took the image statistics as the global item and the image local fitting information as the local item, then combined the above two through the weight parameter to obtain a new Signed Pressure Force (SPF) function integrating global and local information ; afterward, the new SPF function is introduced into the Selective Binary and Gaussian Filtering Regularization Level Set (SBGF-RLS) model so that we constructed a new active contour image segmentation algorithm; finally, we selected four sets of different types of images for MATLAB simulation experiments, and compared the proposed model with LBF Model, CV model and SBGF-RLS model . Experimental results show that the proposed model can quickly and accurately segment images with weak boundaries and uneven gray levels, and has the advantage of being insensitive to initial contours.