Estimation of planar angles from non-orthogonal imaging

Area of Science:

• Computer Vision
• Geometric Measurement

Background:

• Traditional photogrammetry relies on expensive equipment and complex algorithms for 3D scene reconstruction.
• Accurate geometrical parameter estimation is crucial for various engineering and measurement applications.

Purpose of the Study:

• To develop a computationally inexpensive method for estimating planar angles using smartphone imagery.
• To enable on-site geometrical measurements with readily available technology.

Main Methods:

• Utilized two-view geometry and linear algorithms for sparse reconstruction.
• Employed scale-invariant feature transform (SIFT) for point correspondence.
• Applied epipolar constraints and the eight-point algorithm to estimate the essential matrix.
• Determined camera projection matrices and used linear triangulation for 3D point cloud generation.

Main Results:

• Successfully estimated planar angles from non-orthogonal smartphone images.
• Achieved an average error of 3% on non-curved edges.
• Completed estimations in under 10 seconds using only ten images.

Conclusions:

• The developed method offers a simple and efficient approach for geometrical measurements.
• Its integration into smartphones allows for accessible on-site measurements and deformation analysis.