Symmetrical Epipolar Features over Normalized Camera/Projector Calibration Matrices for Real-Time Structured Light Illumination

Kai Liu,Songlin Ying,Daniel L. Lau,Ce Zhu,Bin Xu
DOI: https://doi.org/10.1109/lsp.2022.3217442
2022-01-01
IEEE Signal Processing Letters
Abstract:Structured light illumination is a 3D scanning technique based on projecting a series of striped patterns and reconstructing depth based on the observed warping of the pattern across the target surface. Extensively studied, real-time performance is of paramount importance in many applications. In this Letter, we build on prior researches with epipolar geometry to further simplify the computations of 3D point clouds, with the symmetrical epipolar features derived by extending epipolar geometry over normalized calibration matrices of the camera and projector. Experiments show that the proposed two new processes are of the same accuracy over the normalized calibration matrices with substantially fewer calculations.
What problem does this paper attempt to address?