Wavelength-sensitive-function controlled reflectance reconstruction.

Area of Science:

• Computer Vision
• Computational Imaging
• Color Science

Background:

• Spectral reflectance serves as an object's unique identifier, crucial for applications like color reproduction and computer vision.
• Traditional methods treat all wavelengths equally, neglecting the varying sensitivities of human eyes and imaging sensors.

Purpose of the Study:

• To develop a novel spectral reflectance reconstruction method that accounts for wavelength-dependent sensor sensitivities.
• To enhance the accuracy of reflectance reconstruction, particularly at wavelengths critical for sensor performance.
• To improve performance in downstream applications such as digital imaging and color reproduction.

Main Methods:

• A wavelength-sensitive function (WSF) was derived from sensor-sensitive functions (color matching functions).
• A Hadamard product was used to combine the reflectance matrix with the WSF matrix.
• Singular value decomposition (SVD) was applied to the generated matrix for reflectance reconstruction.

Main Results:

• The proposed method demonstrated a significant reduction in reconstruction errors.
• Mean-square error was reduced by 47% compared to classical methods.
• Lab error was reduced by 55% compared to the principal component analysis (PCA) method.

Conclusions:

• The novel method effectively reconstructs spectral reflectance by incorporating wavelength sensitivity.
• This wavelength-aware approach leads to superior accuracy and performance in imaging and color reproduction tasks.
• The findings offer a more robust solution for spectral reflectance reconstruction in various scientific and industrial applications.