Colour matching by principal component analysis-based spectrophotometric technique

• Published in: Coloration technology 2009-02, Vol.125 (1), p.36-42
• Main Author: Shams-Nateri, A
• Format: Article
• Language: English
• ISSN: 1472-3581; 1478-4408
• DOI: 10.1111/j.1478-4408.2008.00173.x

One of the most important aspects of the physics of colour is colour matching and recipe prediction. In this work, a new method is presented to evaluate the concentration of dyes in textile fabrics. The algorithm of the new technique is based on principal component analysis and single constant spectrophotometric matching methods. Using the normal spectrophotometric matching method, match prediction is carried out at three and 16 wavelengths. With the new algorithm, 3, 6, 9, 12 and 16 principal components are used in spectral match prediction. The performance of the new method improves with increasing numbers of principal components. In addition, the recipe prediction accuracy of the new method with three principal components is better than spectrophotometric matching at three wavelengths. The accuracy of principal component analysis–spectrophotometry with 16 principal components is comparable with the normal spectrophotometric matching method at 16 wavelengths.