Page 42 - Microsoft Word - автореферат_Косаревич.docx

40

процессов для их анализа. Установлено соответствия между характеристиками
первого и второго рода точечных образов и характеристиками макроструктуры
материала, в частности размера зерна и ориентации структурных составляющих
поверхности. Предложенные стохастические модели, описывающие
пространственно распределенные объекты, взаимодействующие, применяемые к
ряду задач, которые возникают при коррозионной деградации материала.
Ключевые слова: сегментация изображений, случайный точечный процесс,
характеристики точечного поля, стохастические модели зарождения и роста
питингов.

Annotation
Kosarevych R.Ja. The development of optical imaging analysis methods of
inhomogeneities on the surface material for monitoring the technical condition of the
objects. - Manuscript.
Thesis for Doctor’s degree in technical science by speciality 05.12.10 - Diagnostic
materials and structures. – Karpenko Physico-Mechanical Institute National Academy of
Science of Ukraine, Lviv, 2016.
The thesis is devoted to the problem of improving the efficiency of optical methods
of control and diagnostics of surface material by selection, quantitative description of the
creation and definition of the parameters of structural irregularities generated during
operation. Based on a common approach - the study of the characteristics of random point
processes developed methods for analyzing metallographic images of technical
diagnostics. All stages of image segmentation and modeling of surface defects based on
the use of random point pattern, their formation, defining their parameters and

characteristics, modeling of development.
The methods of metallographic image segmentation using mathematical models of
various elements of the macrostructure material, due to the terms of the image. For the
first time the method for forming point pattern images based on real images is proposed.
This allows to apply the methods of stochastic point processes for analysis. A new method
of estimation of quantitative parameters of materials parts based on the characteristics of
the first and second type of random point fields is generated. It can be possible to estimate
a grain size for metallographic images obtained in real conditions; check the existence of
surface irregularities as ribbons of carbides, and generate the systems of deterministic
attributes to classify the type of fracture image.
For the first time it was proposed to check the existence and determine the type of
interaction between local corrosion damages of stainless steels based on estimates of the
parameters constructed on the basis of metallographic images stochastic model of the
marked random point process with steam function interaction in the form of Fiksel.
Specified the stochastic model of pitting corrosion of the stainless steel 08Х18Н10Т
based on Gibbs marked random point processes, describing the processes of birth, growth,
and spread on the surface of local corrosion damage and, unlike existing models, taking
into account the interaction between them.
Key words: Image segmentation, random point process, points field characteristics,
stochastic models of nucleation and growth of corrosion pits.