Bi-Polynomial Modeling of Low-Frequency Reflectances

Area of Science:

• Computer Vision
• Computer Graphics
• Material Science

Background:

• Existing reflectance models often prioritize photo-realistic rendering, capturing the complete reflectance domain.
• This can be suboptimal for inverse problems requiring specific frequency components of reflectance.
• High-frequency reflectance details are often noise or irrelevant for tasks like shape estimation.

Purpose of the Study:

• To introduce a novel bi-polynomial reflectance model.
• To accurately represent the low-frequency component of reflectance.
• To facilitate robust solutions for inverse problems in computer vision and graphics.

Main Methods:

• Developed a bi-polynomial function to model surface reflectance.
• The model intentionally discards high-frequency reflectance information.
• Retains essential nonlinear variations within the low-frequency spectrum.

Main Results:

• The bi-polynomial model precisely represents the low-frequency reflectance component.
• Experimental results show superior performance compared to existing parametric models.
• Demonstrated effectiveness in reflectometry and photometric stereo applications.

Conclusions:

• The proposed bi-polynomial reflectance model offers a specialized approach for inverse problems.
• It provides improved accuracy in estimating object reflectance and shape.
• Outperforms traditional models in key computer vision applications.