RESEARCH OF THE MORPHOLOGY OF THE CONTACT ZONE OF COPPER WELDED CONTACTS USING IMAGES SEGMENTATION OF STRUCTURAL ELEMENTS BASED ON WAVELET TRANSFORM

Abstract

The article focuses on the analysis of the morphology of the contact zone of copper junctions using image texture segmentation algorithms and the OpenCV library.

The article discusses the main approaches to the selection of interphase boundaries using the OpenCV graphics library and the use of the Gabor filter to select pore contours, detect defects and pores on interfaces, and establish contact zone phase ratios.

The image processing procedure, based on Gabor filtering, was developed and tested.

The phase areas at the interfaces of the diffusion zones of the Cu-Sn system were calculated for the samples obtained for different modes of copper electrodeposition: stationary, pulse-reverse, and stochastic.

The phase area analysis procedure has been improved. Attempts have been made to apply the texture segmentation procedure to images in which the amplitude of the intensity of gray gradations is proportional to the amplitude of the noise of gray gradations.

An attempt was made to texture the elements of the lamellar structure in the alloy from images of raster electron microscopy.

It was shown that the texture segmentation algorithm, which uses the Gabor wavelet filter, allows the separation of image elements representing defective areas and diffusion contact unnecessary areas.

It was shown that texture segmentation algorithm allows the identification of areas where slight changes in the color intensity gradient (gradations of gray) are possible.

It was shown that usage of bilateral filtering, or Gaussian blur in combination with the Gabor wavelet filter, allows a clearer picture of image boundaries where there is a gradient of gray gradations for low-quality images (images that contain significant color differences in the noise presence).