Spectrum Estimation for Colored Surfaces Using Weighted Pattern Averaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for characterizing color reproduction devices require a large number of patterns to achieve accurate color models, leading to production slowdowns, while reducing the number of patterns results in significant precision loss, and existing hybrid approaches require additional measurements to improve accuracy.
Innovation Solution
A method and device that estimate the spectrum of a colored surface by modifying the spectral components of reference patterns to approximate the desired spectrum, using a subset of patterns and weighted averaging to achieve accurate color reproduction with a reduced number of patterns, typically achieving an average ΔE < 2 with 36 patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of patterns are used for characterization, then color model accuracy is improved, but production speed decreases
Solution Approach 1:
The patent extracts only the essential spectral information from a reduced set of patterns by selecting patterns with diverse spectral characteristics. Instead of using all possible patterns, the method identifies and uses a minimal subset that captures the necessary spectral variations, thereby maintaining accuracy while reducing the number of patterns from over 1500 to approximately 40-100 patterns.
Solution Approach 2:
The patent applies local quality by weighting patterns differently based on their spectral characteristics and importance to the color model. Not all patterns contribute equally; the method identifies which patterns provide the most informative spectral data and assigns higher weights to those, while reducing or eliminating less informative patterns from the characterization set.
2Productivity
If the number of patterns is reduced, then production speed is improved, but color model accuracy deteriorates
Solution Approach 1:
The patent performs preliminary spectral analysis and selection of patterns before the actual color reproduction process. By pre-identifying the most informative patterns and their spectral characteristics, the system prepares a optimized characterization set in advance, allowing accurate color modeling to be achieved with fewer patterns without compromising the speed of subsequent production operations.
Solution Approach 2:
The patent changes the parameters of pattern selection and weighting to optimize the balance between accuracy and speed. By adjusting selection criteria, weighting factors, and the number of patterns used, the system achieves accurate color models (ΔE < 2) with a reduced pattern set, transforming the traditional trade-off relationship.
3Productivity
If a physical approach with few patterns is used, then production speed is maintained, but color model accuracy decreases
Solution Approach 1:
The patent introduces an intermediary computational layer that bridges the physical measurement of few patterns and the required accurate color model. This intermediary involves spectral decomposition, pattern selection algorithms, and weighted averaging computations that transform the limited physical measurements into a comprehensive and accurate color reproduction model, achieving both speed and accuracy.
Data Source
Figure 1~2
Figure 3~4
Figure 5~8
AI summary
Method for determining a reflection spectrum or transmission spectrum of an element of a first surface, said element having been coloured by a device for reproducing colour on the basis of a mixture of several dyes in predetermined proportions, said mixture defining a reproduction hue in respect of the colour of said element, said method comprising: - a step (E60) of selecting a subset of patterns from among a set of reference patterns reproduced or simulated as being reproduced on a second surface exhibiting an identical texture to said first surface, a reflection spectrum or transmission spectrum and a reproduction hue in respect of the colour specific to said device of each of the patterns of said set being known; - for each of said subset, a step (E90) of determining a spectrum adapted at least on the basis of the reflection spectrum or transmission spectrum of said pattern; - a step (E130) of determining said reflection spectrum or transmission spectrum of said element by weighted averaging of said adapted spectra of said patterns of said subset.