A New Two-Dimensional Direction Histogram Based Entropic Thresholding

Abstract

Since the entropic concept was introduced in image segmentation by Pun, many entropic methods have been developed in rapid succession. Consequently, the entropic approach has become a major category of segmentation. Abutaleb et al. firstly introduced the two-dimensional entropies, taking the spatial correlation into account. However, their method takes up too much computational time to give better results. In this paper, a new entropic thresholding method is proposed based on the two-dimensional Shannon entropies by using the orientation histogram. Utilizing several images, the new method is compared with Abutaleb's method and the two-dimensional entropic thresholding method proposed by Xiao et al. The experimental results demonstrate the effectiveness of the proposed method in aspects of two-dimensional entropic thresholding.