A Novel Thresholding Methodology Using WSI EMD and Adaptive Homomorphic Filter

Abstract

In this brief, we propose a novel thresholding methodology using a combination an adaptive homomorphic filter (AHF) and a weighted sinc interpolated empirical mode decomposition (EMD) model for medical and nonuniformly illuminated image datasets. In this methodology, we use an AHF to correct the illumination of an image. Later, we obtain a smooth data component from the histogram of an illumination corrected image using weighted sinc interpolated EMD model. From this smooth data component, we obtain an optimal threshold value using global minimum criteria. By using this optimal threshold value, we obtain a thresholded image from the illumination corrected image. The proposed methodology is very simple and more robust to noise and easy for hardware implementation. The performance of the proposed thresholding methodology using weighted sinc interpolated EMD model is compared with other methods in the literature and the results clearly demonstrate the superiority of the proposed methodology for thresholding.