Abstract. A mathematical model, algorithm, and software are developed for automatic determination of Gaussian noise level on digital images by the method of high-pass filtration. The noise level is calculated for the regions of interest of image, selected by low-pass filtration. The optimal parameters of low-pass and high-pass filters are obtained. Processing a series of test images showed that the proposed method provides the less error of noise level determination than other analogues methods do.

Keywords: digital image processing, determination of noise level, additive white Gaussian noise, convolution, high-pass filtration, region of interest.