SOME PRACTICAL CONCERNS RELATED TO COMPUTER PROCEDURES OF PROCESSING IMAGES IN MATERIAL SCIENCE

The traditional approach to ranking structures and fractures as the comparison with standards (pictures) does not allow to objectively describe the existing diversity of their geometry, to provide the direct comparison of the morphology of structures and fractures to identify critical parameters of structures determining the difference in their resistance to fracture. Formalization of approaches to the description of digital images of structures and fractures is complicated, in particular, due to the differences in the mechanisms of destruction of nominally similar structures that differ in the geometry of the structure of its individual elements and their configuration as a whole; the resulting differences in the metrological support of image processing procedures. It is usually understood that this is provided by default within the framework of the specialized software products used, but in practice, the necessary attention to comparing the alternative options for image processing procedures to choose the optimal one is not always paid. In this regard, the paper considers some aspects of obtaining digital images of structures and fractures, their processing, providing reproducible and comparable results that carry meaningful information about their morphology. In particular, the authors evaluated the role of the etching duration, the choice of the optimal magnification of the microscope (in the range of values comparable in their capabilities to solve a specific problem), and the noise removal procedure. The paper discusses the approaches to the selection of effective image processing algorithms, for example, when changing over from the point estimates of structures to the measuring of their geometry (taking into account the ideas about the statistical nature of structures and fractures, measurement scales). The authors estimated the efficiency of using classical and nonparametric statistics when comparing the results of measurements of the structures and fractures geometry, and the possibility of classifying “blurry” images of microstructures based on the Fourier transform. Based on the results obtained, the authors reviewed some procedures for processing the images of structures and fractures and showed that the use of statistical characteristics of images of structures and fractures makes it possible to rank more objectively the structures according to their geometry.

structure morphology, etching, noise filtering, excess and asymmetry coefficients, field of view area, digital image, structure and fracture images, banding image in microstructure, image processing procedures