Dynamic pupil-segmented polarization imaging using electrically controlled birefringence liquid-crystal for adaptive nanotexture inspection

Machine vision systems for industrial product inspection require precise the characterization of fine surface structures. Differential interference contrast microscopy is widely used to observe surface nanotextures because of its high sensitivity and contrast. However, it does not permit flexible adjustment of the magnitude and directional dependence of the phase sensitivity. Pupil-segmentation techniques overcome this limitation by detecting polarization pattern changes in the pupil plane, which reflects the nanotexture characteristics of the observed surface. This paper introduces an inexpensive, multi-segment, electrically controlled birefringence liquid-crystal element to dynamically optimize segmentation patterns. By individually adjusting the birefringence phase difference of each segment using the applied voltage, we achieved real-time pupil-segmentation optimization. This enables flexible and adaptive detection of nanotextures with different orientations and contrasts within a coaxial illumination system. Experiments demonstrated that dynamically optimizing pupil-segmentation patterns enables the adaptive control of both directional sensitivity and detection capability for various surface nanotextures and step structures. This capability is expected to facilitate cost-effective, high-speed, and high-precision surface inspection across various industrial applications, such as the selective observation of directionally dependent surface steps, suppression of uniform processing marks, and detection of fabrication-induced scratches on highly reflective materials, including metals.