High-precision tunable three-dimensional imaging based on meta-device

Structured-light three-dimensional (3D) reconstruction has found broad applications in fields such as machine vision, plastic surgery, and cultural heritage preservation. In recent years, metasurface-based structured-light projection has created new opportunities for high-precision 3D measurement, due to its advantages of large depth of field, wide field of view, and compact design. However, once the metasurface is fabricated, its spatial sampling rate becomes fixed, fundamentally limiting the reconstruction resolution. To address this challenge, we incorporated liquid lenses into the optical path on the hardware side, enabling dynamic adjustment of the spatial sampling rate. On the algorithmic side, by combining sub-pixel circle-center extraction with geometric constraints, the matching process becomes primarily reliant on localization accuracy, thereby eliminating dependence on conventional region-based correlation matching and significantly improving both reconstruction robustness and accuracy. Experimental results demonstrate that the proposed method achieves a 17 times improvement in spatial resolution over traditional approaches while maintaining system compactness and providing a depth accuracy of 0.0349 mm. This work provides a novel technical pathway for real-time, high-precision 3D reconstruction in applications such as facial recognition and wearable devices.