Prediction of DCT-based denoising efficiency for images corrupted by signal-dependent noise

This paper describes a simple and fast way to predict efficiency of DCT-based filtering of images corrupted by signal dependent noise as this often happens for hyperspectral and radar remote sensing. Such prediction allows deciding in automatic way is it worth applying denoising to a given image under condition that parameters of signal-dependent noise are known a priori or pre-estimated with appropriate accuracy. It is shown that denoising efficiency can be predicted not only in terms of traditional quality criteria as output MSE or PSNR but also, with slightly less accuracy, in terms of visual quality metrics and PSNR-HVS-M.