Automation of Measuring Microhardness of Materials using Metal-Graphic Images

The research is devoted to the development of a system that combines classical methods of digital image processing and the use of convolutional neural networks to solve the problem of calculating the microhardness of metals from digital metallographic images. The article describes the design of the developed module of prism prints, the geometrical characteristics of which are characterized by microhardness. The article presents the results of the study of the developed system on real samples in the laboratory. The analysis of the results showed that the separate use of classical methods of digital image processing and neural networks does not provide a good result. Therefore, it is preferable to combine approaches to solve the problem.