Abstract

For the gradient image threshold segmentation, threshold selection is not fully take into account the boundary features in the edge function of MATLAB. By analyzing high gradient magnitude or local bigger change of gradient image boundary region, the transition region extraction algorithm is present. The gradient and entropy is studied, and fusion algorithm of them is designed. The total transition region is composed of pixels with high gradient magnitude and the larger local entropy, then threshold selected based on transition region histogram, the boundary segmentation can be realized. Experiments show that the algorithm is effective and practical.

Paper Details