UV photogrammetry for transparent and composite surfaces

Abstract

Transparent and composite surfaces pose a fundamental challenge for stereo photogrammetry: optically smooth glass produces no detectable surface features under visible illumination, making three-dimensional reconstruction impossible without surface preparation. This excludes optical components such as lenses and cover glasses, composite assemblies, and semi-translucent biological specimens from non-contact geometric measurement. Here we show that coherent speckle illumination at 266 nm overcomes this limitation by exploiting wavelength-dependent scatter enhancement, generating sufficient backscattered signal on surfaces that are entirely invisible under visible illumination. We developed a multispectral stereo system and evaluated three illumination modalities under identical acquisition conditions. On transparent glass, both visible modalities produce complete reconstruction failure, recovering only non-transparent holder structures. Ultraviolet speckle illumination at 266 nm enables dense reconstruction of the same surfaces. We demonstrate recovery of an uncoated plano-convex lens with a fitted radius of 30.946 mm and point-cloud standard deviation of 106.5 {micro}m, defect detection on a transparent cover glass without surface preparation, and reconstruction of a semi-translucent biological specimen. On metrology-grade reference objects, ultraviolet speckle achieves a standard deviation of 116 {micro}m and completeness exceeding 93%, approaching the performance of optimised visible structured illumination. These results establish ultraviolet speckle photogrammetry as an enabling approach of optical metrology to otherwise uncooperative surfaces, with relevance to optical manufacturing inspection and biological surface analysis.