Improved multi-input parameter optimization method for camera colorimetric characterization

Area of Science:

• Color Science
• Image Processing
• Computer Vision

Background:

• Accurate camera colorimetric characterization is crucial for reliable image reproduction.
• Traditional methods often rely solely on RGB values, limiting their precision.

Purpose of the Study:

• To propose and validate a multi-input parameter optimization method for improving camera colorimetric characterization.
• To introduce luminance (L) as an additional input parameter alongside RGB, creating an RGBL model.

Main Methods:

• Developed an RGBL four-input parameter conversion model using polynomial and backpropagation (BP) neural network approaches.
• Utilized a spectroradiometer and three cameras to measure RGBL and CIEXYZ values for 549 Munsell colors.
• Compared the RGBL model's performance against traditional three-input parameter models.

Main Results:

• The proposed RGBL method significantly enhanced conversion accuracy.
• A substantial reduction in color difference, up to 57.7% in CIELAB, was achieved.
• The RGBL model demonstrated superior performance over RGB-only models.

Conclusions:

• Introducing luminance (L) alongside RGB values (RGBL) improves camera colorimetric characterization.
• The RGBL method offers a more accurate and reliable approach to color conversion in digital imaging.