Shape and Reflectance Reconstruction Using Concentric Multi-Spectral Light Field

Area of Science:

• Computer Vision
• Optical Engineering
• Material Science

Background:

• Recovering shape and reflectance of non-Lambertian surfaces is difficult due to view-dependent appearance.
• Traditional photo-consistency constraints fail for these surfaces.

Purpose of the Study:

• Introduce a novel concentric multi-spectral light field (CMSLF) system.
• Enable single-shot recovery of shape and reflectance for diverse materials.
• Improve depth estimation for specular points.

Main Methods:

• Designed a CMSLF system with concentric camera rings capturing specific spectra.
• Utilized a multi-spectral ring light for spectral multiplexing of viewpoint and lighting variations.
• Formulated a physics-based reflectance model for CMSLF under the surface camera (S-Cam) representation.

Main Results:

• The CMSLF system enables single-shot shape and reflectance recovery.
• A unique single-peak specularity pattern aids robust depth estimation for specular points.
• Outperformed state-of-the-art shape reconstruction methods in extensive experiments.

Conclusions:

• The proposed CMSLF system effectively recovers shape and reflectance from non-Lambertian surfaces in one shot.
• The method shows significant improvements over existing techniques, particularly for challenging materials.
• The unique camera and light source configuration facilitates robust specular point depth estimation.