Defect enhancement method for ZrO2-bearing spherical surface representation based on complete defect stitching and self-defined background balancing algorithm

To address the problems of difficulty in extracting complete defects and high interference of surface background in the process of ZrO2 bearing ball defect detection. A method of ZrO2 bearing ball surface representation defect enhancement based on complete defect stitching and self-defined background balance algorithm is proposed. To determine the location of complete defects, we propose a scanning and acquisition scheme for sub-aperture images and quickly stitch the edges of sub-aperture images. The image feature point gradient definition is optimized, and a less error Euclidean distance finding method is used for feature point matching. The stitched complete defect image is smooth and has high pixel accuracy. The background feature layer of the complete defect image is extracted for background gray leveling, and this process is introduced into the γ-value calculation process. The γ setting allows the image to be self-defined enhancement, outputting a fully defective image with a balanced background and outstanding defects. The average gradient is reduced by more than 90% after complete defect image enhancement, and the information entropy is preserved by about 40%. The defect edges of the image are intact, and the fine texture of the defect features is preserved.