Color correction for printing devices with the use of neural networks

摘要

Color images reproduced from image files by electronic devices always have some color inaccuracy, which is attributed to many factors. For example, in a printed image it may be caused by the type of paper, the error in converting the RGB color representation into the CMYK representation, imprecision from the printer, etc. Correction after scanning and before printing is necessary in order to improve the color quality. In this study, we investigated the neural network technology for this application. CMAC-based neural networks have been selected for implementing the mapping for converting the original color command to a modified one for a printing device. Results show that significant improvement can be achieved.