Predicting the development of an observed process from its digital image

The purpose of the study is to develop a method for predicting the development of observed processes of various natures using their digital images containing information that is inaccessible to visual perception. The scientific novelty lies in the transformation of digital images, which allows us to qualitatively and quantitatively assess the state of the observed process (system). In this case, a qualitative analysis of the dynamics of changes in a digital image is carried out visually with an accuracy of one pixel. To quantify the stability of the system, the Hurwitz criterion was used. The developed method was used to assess the surface strength of fatigue specimens under bending loading. As a result, it was found that small changes on the surface of the specimen, which are not visually recorded, make it possible to determine the moment when the system starts to lose its steady state.