[Sequential selection of representative color samples for spectral reflectance reconstruction].

Area of Science:

• Color science
• Image processing
• Spectroscopy

Context:

• Spectral reflectance reconstruction is crucial for accurately characterizing object surfaces using color charts.
• Current methods for selecting representative colors from charts are time-consuming and may not be optimal for specific imaging systems.
• Existing techniques focus on spectral reflectance distribution, neglecting system-specific requirements.

Purpose:

• To propose a novel sequential method for selecting the most representative colors for spectral characterization.
• To optimize color sample selection for improved spectral and colorimetric accuracy in imaging systems.
• To address the limitations of existing methods that do not account for specific imaging system characteristics.

Summary:

• A two-step sequential method is presented for selecting representative colors.
• The first step selects initial colors minimizing spectral root-mean-square error using a virtual system.
• The second step refines selection based on the real imaging system's spectral responsivity, enhancing accuracy.

Impact:

• The proposed method significantly improves both spectral and colorimetric accuracy compared to previous techniques.
• Demonstrated effectiveness on diverse systems: an 11-channel multispectral imager and a 3-channel color scanner.
• Enables more efficient and accurate spectral characterization in practical applications.