Photometric Stereo in Participating Media Using an Analytical Solution for Shape-Dependent Forward Scatter

Area of Science:

• Computer Vision
• Optics
• Image Processing

Background:

• Participating media (e.g., fog, smoke, murky water) scatter light, degrading image quality.
• Traditional 3D reconstruction methods struggle with scattered light, limiting their application in such environments.

Purpose of the Study:

• To develop a novel photometric stereo method capable of accurate 3D reconstruction in participating media.
• To address the limitations of existing methods by effectively modeling and mitigating light scattering effects.

Main Methods:

• Proposed a photometric stereo method specifically designed for participating media.
• Introduced a new model for shape-dependent forward light scatter using analytical forms and spatially-variant kernels.
• Developed an approximation of dense matrices as sparse matrices to facilitate the removal of forward scatter.

Main Results:

• The proposed model accurately describes forward scatter using lookup tables and spatially-variant kernels.
• The sparse matrix approximation effectively enables the removal of forward scatter.
• Experiments with both synthesized and real data demonstrated significant improvements in 3D reconstruction quality.

Conclusions:

• The novel photometric stereo method enhances 3D reconstruction accuracy in participating media.
• The approach effectively models and removes forward light scatter, overcoming limitations of traditional techniques.
• This work opens new possibilities for 3D reconstruction in visually degraded environments.