Reflectance and Fluorescence Spectral Recovery via Actively Lit RGB Images

Area of Science:

• Computer Vision
• Computational Imaging
• Optical Physics

Background:

• Fluorescence analysis in computer vision offers valuable scene information for applications like spectral sensitivity estimation and 3D reconstruction.
• Traditional hyperspectral imaging of fluorescent scenes requires specialized, costly equipment (hyperspectral cameras, illuminants).

Purpose of the Study:

• To develop a practical, cost-effective hyperspectral imaging method for reflective-fluorescent scenes.
• To enable accurate spectral estimation and scene relighting using conventional hardware.

Main Methods:

• Utilizing a standard RGB camera and varied colored illuminants.
• Exploiting the illuminant-invariant chromaticity of fluorescent emissions.
• Estimating spectral reflectance and fluorescent emission chromaticity.
• Deriving fluorescence absorption and emission spectra.

Main Results:

• Accurate estimation of all scene spectra from RGB images.
• Robust estimation of spectral reflectance and fluorescent chromaticity.
• Successful relighting of scenes under novel lighting conditions.

Conclusions:

• The proposed method provides a practical alternative to specialized hyperspectral cameras for fluorescent scene analysis.
• Accurate spectral and fluorescence properties can be recovered using conventional RGB cameras and controlled illumination.
• The technique facilitates advanced applications like scene relighting with improved accuracy and accessibility.