There exist some shallow scratch defects on the super-smooth optical surface. Their detection has a low efficiency with the existing technologies. So a new detection method, dark-field detection of adaptive smoothing and morphological differencing (DFD-ASMD), is proposed. On one hand, the information of shallow scratches can be kept in dark-field images. On the other hand, their weak characteristics can be separated and protected from being overly reduced during the elimination of noise and background in the image. Experiments show the detection rate of shallow scratches is around 82%, and DFD-ASMD can lay a foundation for quality control of defects on the high-quality optical surface.

A surface defect evaluation system can combine microscopic scattering dark-field imaging with sub-aperture scanning and stitching. Thousands of sub-apertures are involved; mechanical errors will cause stitching dislocation, leading to defect cracks. In this Letter, we propose standard line coordinate error adjustment dealing with consistency error between coordinates of the scanning and imaging system, and defocus depth estimation leveling method dealing with high-cleanliness fine optics defocuing caused by the surface which is not perpendicular to microscope’s optical axis. Experiments show defect cracks are effectively solved and the defocus of 420 mm×420 mm components can be controlled within depth of field 20 μm.