Nondestructive Detection Algorithm for Si₃N₄ Bearing Roller Microcrack Characterization Based on Multiscale Gamma Correction and Growth Region Segmentation

For the Si₃N₄ bearing roller microcracks micro-motion wear characterization, such as low contrast, small pixel brightness value ratio, and unclear detail texture structure. A fusion algorithm of multiscale Gamma correction and growth region segmentation is designed, and the precise detection and extraction of the Si₃N₄ bearing roller microcracks micro-motion wear characterization is realized. A multiscale Gamma correction framework is constructed to perform gray value nonlinear operations in both frequency and time domains. This framework enhances the contrast and pixel brightness of microcrack images. To solve the problem of unclear detail texture structure in the image, a seed point selection method is designed to combine the Sobel operator and gradient amplitude value. The aim is to achieve growth segmentation of pixels inside the defect, while effectively preserving the detail texture structure of micro-motion wear characterization. The average gradient value of the enhanced Si₃N₄ bearing roller microcracks micro-motion wear characterization has an average growth rate of 98.8%. The average accuracy of the image extracted from the Si₃N₄ bearing roller microcracks micro-motion wear characterization reaches 83.92%. The method effectively preserves the detail texture and structural characterization in microcracks, and accurate nondestructive detection of the Si₃N₄ bearing roller microcracks micro-motion wear characterization is achieved.