The optimum automatic thresholding using the phase of Zernike moments

A new technique for automatic thresholding of images has been introduced. This technique is based on maximizing the correlation between Zernike moments´ phases of the gray-level and binary images of the same objects. This technique of gray level thresholding is unimodal. Thresholding using Zernike moments would be of interest to pattern recognition applications where Zernike moments are used as features. The experimental results show that correlating the phases of Zernike moments yields the optimal threshold values. These results also indicate the robustness and stability of the technique when dealing with noisy sample images.