Adaptive spectral characterization model for various types of displays

Area of Science:

• Display technology
• Color science
• Image processing

Background:

• Accurate display characterization is crucial for color reproduction.
• Existing models often require extensive training data.
• Limited data scenarios pose challenges for spectral characterization.

Purpose of the Study:

• To develop an adaptive spectral characterization model for diverse displays.
• To improve prediction accuracy with limited training data.
• To enable efficient and accurate color and chromaticity characterization.

Main Methods:

• An adaptive spectral characterization model with characteristic fitting and crosstalk compensation modules.
• Iterative training of the characteristic fitting module to learn tone reproduction curves.
• Oblique cubic sampling method for efficient data acquisition.

Main Results:

• Excellent performance across five display types with only 128 training samples.
• Average spectral root mean square error below 10^-4.
• Average CIEDE2000 color difference below 1.2.

Conclusions:

• The proposed model offers high accuracy and efficiency for spectral characterization.
• Adaptive learning and efficient sampling enable performance with limited data.
• The model's adaptability extends to chromaticity characterization for specific applications.