Large-depth-range 2D and 3D reconstruction based on unidirectional parallel single-pixel imaging

Area of Science:

• Optics and Photonics
• Computer Vision and Image Processing
• Metrology

Background:

• Traditional structured-light 3D reconstruction methods suffer from significant measurement errors outside the depth of field (DOF) due to defocus blur.
• This limitation restricts their application in scenarios requiring large working volumes.

Purpose of the Study:

• To propose a novel unidirectional parallel single-pixel imaging (U-PSI) method for simultaneous 2D image deblurring and accurate 3D measurement across a large depth range.
• To overcome the DOF limitations inherent in conventional 3D reconstruction techniques.

Main Methods:

• The U-PSI method projects unidirectional Fourier-basis patterns.
• It derives the 2D point spread function (PSF) from the 1D light transport coefficient (LTC) using PSF-LTC reciprocity for deblurring.
• It leverages the centroid position invariance to defocus of 1D LTC for consistent large-depth 3D reconstruction.

Main Results:

• The proposed U-PSI method achieves simultaneous accurate 2D imaging and 3D measurement.
• The system demonstrated effective performance within a substantial depth range of 360 mm.
• Accurate 3D reconstruction was maintained despite defocus blur.

Conclusions:

• The U-PSI method offers a robust solution for accurate 3D measurement in large-scale applications.
• It shows significant promise for fields like industrial inspection, cultural heritage digitization, and architectural surveying.
• This technique expands the practical applicability of 3D imaging in extended working volumes.