A saturation channel detection method for surface defects of silicon nitride bearing rollers based on adaptive gamma correction-edge threshold segmentation coupling algorithm

Abstract

For the problem that the surface defects in Si3N4 ceramic bearing rollers are fuzzy, it is difficult for traditional single algorithms to achieve accurate localization and detection of surface defects. An adaptive gamma correction and edge threshold segmentation coupling algorithm is designed to accurately detect surface defects in the intensity channel of Si3N4 ceramic bearing rollers. To preserve the surface defects information to the greatest extent, the HSI intensity channel image characteristics of a single surface defects in Si3N4 ceramic bearing roller is analyzed. To obtain the best γ value of a single image, an adaptive gamma correction function equation is constructed. Analyze the frequency-domain characteristics of the three-dimensional grayscale curve of surface defects, and the grayscale gradient of surface defects is extended. Use the non-maximum value suppression model to suppress noise and background information, and perform sharpening enhancement and threshold segmentation on defect edge information. The mean PSNR of the surface defects images of the Si3N4 ceramic bearing roller is 28.9161 dB. The mean information entropy is 3.7572 bit. The accuracy of the detection result is 96.3%. The surface defects intensity channel detection method effectively improves the accuracy and efficiency of locating and detecting surface defects of Si3N4 ceramic bearing rollers.