Light source position and reflectance estimation from a single view without the distant illumination assumption.

Area of Science:

• Computer Vision
• Computer Graphics
• Material Science

Background:

• Existing methods for surface reflectance recovery often assume distant light sources, limiting their application in scenarios like indoor environments.
• Accurate modeling of surface reflectance and illumination is essential for realistic computer graphics and material analysis.

Purpose of the Study:

• To develop methods for estimating surface reflectance properties and light source position from a single view without the distant illumination assumption.
• To enable the generation of synthetic images under arbitrary lighting conditions, including modifications to source direction and distance.

Main Methods:

• The first method estimates light source position by fitting the Lambertian diffuse component and separating specular/diffuse components using iterative relaxation.
• The second method uses a polarization-derived specular component image to simultaneously recover reflectance properties and light source positions by optimizing a log-transformed Torrance-Sparrow model.

Main Results:

• The proposed methods accurately estimate surface reflectance and light source position from a single image and 3D model.
• The framework successfully relaxes the distant illumination assumption, broadening applicability to complex environments.

Conclusions:

• The developed estimation framework provides a robust solution for surface reflectance and light source localization.
• This work advances the capabilities for realistic synthetic image generation and material property analysis under unconstrained lighting.