[Estimation of spectral reflectance using color scanner].

Area of Science:

• Color science
• Image processing
• Spectrophotometry

Context:

• Color scanners often deviate from linear models.
• Accurate spectral characterization is crucial for device-independent color.
• Training data distribution significantly impacts color estimation.

Purpose:

• Propose a novel spectral characterization method for color scanners.
• Improve the accuracy of estimating high-dimensional spectral reflectance from low-dimensional scanner responses.
• Address the limitations of existing color scanner characterization techniques.

Summary:

• A new method estimates spectral reflectance using limited sample selection and weighted training.
• This approach accounts for scanner non-linearity and training data effects.
• The method accurately reconstructs spectral reflectance from scanner responses.

Impact:

• Achieves superior performance compared to previous methods.
• Reduces both color difference error and spectral reflectance error.
• Enables more accurate and reliable color reproduction in scanned images.