Variational mode decomposition based image denoising using semi-adaptive conductance function inspired diffusion filtering

Abstract

In day-to-day life, images are the most frequent and casual way of information sharing. These images are susceptible to external disturbances or noise. Thus, to curb noise, image denoising algorithms are utilized. In this paper, the variational mode decomposition, with its concurrent and a non-recursive process for determining the mode functions that also provides a robust method for image denoising, has been introduced. This decomposition process divides the whole spectrum of the signal into a number of sub-bands or mode functions, centered around their respective center frequencies. To these mode functions, spatial filters such as bilateral filter, wiener filter, and modified anisotropic diffusion filter are employed. These filters help in enhancing the yield of the quality assessment metrics; such as mean square error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM), together with the semi-adaptive conductance function in the diffusion filter. The parameters of these respective spatial filters are calibrated, and then selected in order to get the best possible metric scores. The applicability and ability of the algorithm to suppress noise are compared visually and quantitatively for the noisy image using modified variational mode decomposition and other denoising algorithms in both low and high noise levels. The algorithm provides an average decrease of 62% in case of MSE, 28% increase in PSNR, and 110% increase in SSIM when compared with other denoising techniques. The estimated metric score values signify that the proposed method has a better prospect as a denoising algorithm. © 2023, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.