browse.title.page

Зараз показуємо 1 - 1 з 1

System of automated detection of ceramic disc surface defects
(2021) Konechnyi, B.
handheld USB digital microscope is considered as an instrument for ceramic disk inspection automatization in the production cycle. It provides the opportunity for surface defect detection in the needed range from 50 to 500 μm. Such a device providing 15 fps with the standard VGA resolution of 640x480 pixels enables one track scan for 18 seconds. As a decision unit, a system based on the artificial neural network was used. The basic system software has been developed. A study of the object illumination and the size of the database influence on artificial neural network training results (probability of defect detection) have been carried out. The use of digital filtration, adaptive histogram equalization, threshold function, and feature detection as a classical approach showed a good result. Deep learning has proven its effectiveness, selecting features of the defect and presenting more stable results in defect detection.