Scanner Approximation of In-Line Spectrophotometer in Duplex Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Duplex printing devices face high costs and reduced accuracy due to the need for separate in-line spectrophotometers for each side, and existing machine learning-based methods provide inadequate spectral estimation for a wide range of papers and inks, leading to noticeable color differences.
Innovation Solution
A color management system that uses a reference in-line spectrophotometer for one side and a scanner for the other, with a processor applying corrections based on color data differences to ensure balanced color adjustment and high image quality, eliminating the need for a spectrophotometer on the second side.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an in-line spectrophotometer is used for each side of a duplex printing device, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses a scanner to capture color data from printed patches and creates a spectral estimate as a copy of the actual spectral data. This scanned color data is then processed through machine learning models to generate approximate spectral information, replacing the need for a physical spectrophotometer on the second side while maintaining adequate measurement capability
Solution Approach 2:
The patent replaces expensive, complex spectrophotometer hardware with a more economical scanner-based system. The scanner is a lower-cost device that, when combined with computational algorithms, provides sufficient spectral estimation capability for color management purposes without requiring the precision instrumentation of a dedicated spectrophotometer on each side
2Device complexity
If a scanner is used to estimate spectral data, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent introduces machine learning models as an intermediary between the scanner's raw color data and the required spectral information. These computational models act as a bridge, translating limited RGB or CMYK scanner data into estimated spectral profiles by learning the relationship between colorimetric data and spectral characteristics from training sets
Solution Approach 2:
The patent transforms the scanner's output parameters (RGB or CMYK values) into spectral estimates by changing the parameter representation. Through machine learning transformations, the system converts three-channel color data into multi-wavelength spectral data, effectively expanding the information content through computational parameter transformation rather than physical measurement expansion
3Measurement precision
If machine learning-based training methods are used, then measurement precision is improved, but adaptability deteriorates
Solution Approach 1:
The patent develops machine learning models trained on diverse datasets encompassing multiple paper types and ink formulations. The training process uses varied reference spectral data and corresponding scanner measurements across different media, creating a universal model that can handle various paper-ink combinations without requiring separate specialized models for each material type
Solution Approach 2:
The patent performs preliminary training of the machine learning model using extensive reference data from multiple paper and ink combinations before actual color management operations. This pre-training phase establishes the model's ability to generalize across different materials, and the trained model can then be applied universally to various paper-ink combinations without requiring retraining for each specific combination
Data Source
AI summary
A color management system acquires first color data for a color patch of a test pattern printed by a first image forming device on a recto side of a sheet, acquires second color data for the color patch in the test pattern printed by the first image forming device; and acquires third color data for a corresponding color patch in the test pattern printed by a second image forming device on a recto side of a sheet which has passed through the first image forming device. The system applies a correction to the third color data, which is a function of a difference between the first color data and the second color data and which may also be a function of a difference between the second color data and the third color data.