Sub-aperture image stitching of cylindrical mirror surface defects based on dark-field imaging

High-precision cylindrical lens optics are widely used in high-precision equipment, such as high-power laser resonators and long-distance line interferometers. In most instances, this lens required to possess extremely low surface roughness and surface defects. Using dark-field microscopy to detect the surface defects of the cylindrical mirror is simpler than the interferometric, and the cost is lower. In the experiment, we designed a set of image acquisition devices based on the characteristics of the cylindrical mirrors, and used this device to obtain overlapping sub-aperture images of the cylindrical mirrors. According to the overlapping area to solve the relative position relationship of two adjacent sub-apertures, an image stitching algorithm based on defect location is proposed, which realizes the stitching of multiple sub-apertures. The experimental system was used to scan a plano-convex cylindrical mirror with a size of 22mm×22mm and a radius of curvature of 41.34mm. Splicing the obtained sub-aperture images of surface defects. Experiments show that the stitching algorithm has high matching accuracy and strong robustness.