An on-line color defect detection method for printed matter based on snapshot multispectral camera

As the development of the polychrome printing technology, more and more pigments are available on printing and packaging industry, which has brought new requirements to the on-line color defect detection for printed matter. There are always difficulties for traditional detecting approaches with commercial RGB cameras to provide competent color resolution due to the color gamut limitation. In this communication, we proposed a snapshot multispectral imaging method using a novel spectral filter array (SFA), which has eight spectral channels and one panchromatic channel. Spectral reconstruction and color reproduction was carried out by using BP network with the training on Munsell colors and typical printed samples. We defined the empirical threshold values for color defect detection in the spectral vector space, and demonstrated the validity of this method with practical printed matter experiments.