Development of Methods for Image Filtering in Noise and their Implementation for a Web Service

Abstract Known image processing software systems use filtering methods such as the Gaussian filter, the median filter, and others, which often do not have satisfactory performance in processing certain types of noise. This leads to a partial loss of the useful signal and a deterioration in image quality. The work is devoted to improving the quality of image filtering in various kinds of noise. A model of additive interaction of signals in additive impulse noise is proposed. A new method of least finite differences (MLFD) and its modification for image filtering has been developed. An interactive web service for the implementation of MLFD was created. Various filtering methods have been proposed to reduce the effect of impulsive noise on images. The proposed data processing is based on a priori information about the type of noise. The web service is a flexible and efficient solution for filtering digital images and is not available in well-known graphics packages. The new web service is a good compromise between the quality of filtering and the simplicity of the solution compared to complex and resource-intensive systems for graphic image processing.