A Simple Way to Reduce 3D Model Deformation in Smartphone Photogrammetry

Area of Science:

• Geomatics Engineering
• Computer Vision
• Digital Imaging

Background:

• Smartphone photogrammetry is emerging as an accessible measurement tool.
• Its suitability for Structure from Motion-Multi Stereo View (SfM-MVS) with self-calibration requires investigation.
• Comparison with Uncrewed Aerial Vehicle (UAV) photogrammetry practices is relevant.

Purpose of the Study:

• To assess the geometric stability of smartphone cameras for photogrammetry.
• To evaluate the effectiveness of SfM-MVS with self-calibration versus pre-calibration using smartphone imagery.
• To determine the accuracy and limitations of smartphone photogrammetry for 3D model generation.

Main Methods:

• Geometric stability testing of 14 smartphone cameras using a checkerboard test field.
• Calibration of internal orientation parameters, repeated multiple times.
• Generation of 3D models of a sculpture using SfM-MVS with self-calibration and pre-calibration variants.
• Comparison of smartphone-generated models against a reference model created with a Digital Single-Lens Reflex (DSLR) camera.

Main Results:

• Smartphone cameras demonstrated lower internal orientation parameter stability compared to DSLR cameras.
• Principal distance and principal point position were found to be unstable in smartphone cameras.
• Pre-calibration using test field data improved 3D model geometry over self-calibration, reducing local deformations.

Conclusions:

• Smartphone photogrammetry has significant potential for widespread use.
• The inherent instability of smartphone camera parameters necessitates careful calibration.
• Pre-calibration is recommended over self-calibration for improved geometric accuracy in smartphone photogrammetry applications.