Page 9 - КЛАСИФІКАЦІЯ ЛОКАЛІЗОВАНИХ ДЕФЕКТІВ ЗВАРНИХ ШВІВ НА РЕНТГЕНІВСЬКИХ ЗОБРАЖЕННЯХ ТРУБ

9

alloy with silicon carbide particles, evaluation of porosity of metal coatings,

improving the accuracy of tracking motion in electron beam welding, analysis of

fractographic images of fractures of heat-resistant steels.

The developed methodology is based on the following results.

The current state of the problem of creating methods of image processing and

analysis in non-destructive testing and technical diagnostics is examined. The

analysis of methods of enhancement of digital radiographic images, methods of

localization and classification of the selected defects of welded joints, methods of

segmentation of color images in diagnostic tasks is carried out.

Analysis of publications shows that noise-suppression methods of image

processing are needed. Their use should show the reduction of noise exposure,

uneven lighting in optical images, to provide high-quality radiography of structural

elements with significantly different thickness areas.

A number of methods for improving digital radiographic images is proposed.

The conditions for the optimal expansion of the dynamic range of gray levels of the

image using a generalized model of logarithmic image processing are established.

This improves image quality.

A structure-adaptive method of image enhancement based on a logarithmic

image representation model has been developed. Using a combination of logarithmic
multiplicative operators allows to emphasize the low-contrast boundaries of image

fragments. By choosing the values of the logarithmic model parameter, one can
significantly improve the image quality.

A method for increasing the contrast of X-ray images based on a nonlinear
modification of the detailed coefficients of the wavelet decomposition is proposed. It

allows to increase the generalized absolute contrast of the image compared to other

wavelet methods.

A method for estimating the blurring function X-ray images of welds is

proposed. The reference image method is used to restore them.

The use of a frame diamond-shaped aperture for the median filter reduces the

effect of quantization noise and film grain. With the use of non-sharp masking and