Phase-based optimal image thresholding

The. automatic binarization of gray-level images or the automatic determination of an optimum threshold value that separates objects from their background is still a difficult and challenging problem in many image processing applications. The difficulty may arise due to a number of factors, including, poor contrast, high noise to signal ratio, complex patterns, and/or variable modalities in the gray-scale histograms. In this paper an algorithm for determining an optimum image thresholding value is proposed. Phase correlation between the gray-level image and its binary counterpart is defined as a function of the thresholding parameter. The optimum thresholding problem is then constructed as a problem of optimization where the objective is to find a threshold value that maximizes the phase correlation between the two images. Experimental results to compare the proposed algorithm to the various thresholding techniques are also presented. (C) 2003 Published by Elsevier Inc.