Multispectral imaging colorimetric measurement method with illumination-adaptive white calibration.

Area of Science:

• Color Science
• Optical Engineering
• Computer Vision

Background:

• Accurate color measurement of non-planar surfaces is challenging due to uncontrolled lighting and surface scattering.
• Existing non-contact methods struggle with complex geometries and variable illumination conditions.

Purpose of the Study:

• To develop a multispectral imaging colorimetric measurement method for non-planar surfaces.
• To improve the accuracy and standardization of non-contact color measurements.
• To create a device capable of uniform illumination and adaptive white calibration.

Main Methods:

• Spectral reconstruction using a backpropagation neural network.
• Illumination-adaptive white calibration and spectral fusion techniques.
• A custom device with an integrating sphere, an 8-channel multispectral camera, and a rotating table coated with PTFE for calibration.

Main Results:

• Reduced average reconstruction error from 2.4 to 1.5 CIEDE2000 color difference units after PTFE data correction.
• Maintained color measurement deviation (ΔE00) below 2 across various angles for glossy curved surfaces.
• Demonstrated superior accuracy compared to conventional point light sources for glossy objects.

Conclusions:

• The proposed method provides uniform, stable diffuse lighting for precise non-contact color analysis.
• The integration of digital imaging with spectroscopy offers a powerful tool for accurate color measurement.
• The developed device and method are effective for measuring complex objects like porcelain exhibits.