Color Prediction Model Creation Using Selective Ink Amount Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As the number of ink colors used in printers increases, the number of combinations of ink amount sets becomes enormous, making it impractical to improve color prediction accuracy by increasing the number of samplings, and existing technologies lack efficient methods to enhance accuracy without increasing sampling complexity.
Innovation Solution
A method is introduced to create teacher data for a color prediction model by selecting ink amount sets from a lattice point space using a predetermined selection rule, acquiring spectral reflectance, and setting it as input and output values, allowing for improved color prediction accuracy while reducing the number of samplings. This involves a color prediction model creation apparatus that performs machine learning using this teacher data to predict spectral reflectance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of sampling ink amount sets is increased to improve color prediction accuracy, then color prediction accuracy is improved, but the complexity and cost of measurement and data creation increases significantly
Solution Approach 1:
The ink amount space is divided into multiple regions based on color gamut characteristics (e.g., high lightness region, low lightness region, saturated color region). Sampling is performed selectively in each region according to predetermined criteria, rather than uniformly across the entire space. This segmentation allows focused sampling in critical regions while reducing overall sampling burden.
Solution Approach 2:
Different sampling strategies are applied to different regions of the ink amount space. For example, regions with high color gamut importance or low lightness (where prediction is more difficult) receive denser sampling, while other regions use sparser sampling. This local differentiation optimizes prediction accuracy where it matters most without uniformly increasing sampling complexity everywhere.
2Adaptability or versatility
If the number of ink colors used in the printer is increased to expand color gamut, then color reproduction capability is improved, but the number of ink amount set combinations becomes enormous making sampling impractical
Solution Approach 1:
From the enormous set of all possible ink amount combinations, the invention extracts only the representative combinations that are most important for color prediction. This is done by identifying key regions in the ink amount space (such as boundaries of color gamut, regions with extreme ink ratios, or areas with significant color differences) and selecting samples from these regions, rather than attempting to measure all possible combinations.
Solution Approach 2:
Instead of performing complete sampling of all ink amount combinations (which would be excessive and impractical), the invention performs partial sampling focused on the most critical regions. The sampling is designed to capture the essential color behavior patterns without requiring exhaustive measurement of every possible combination, achieving sufficient accuracy with reduced effort.
Data Source
AI summary
A method of creating teacher data used for creating a color prediction model that predicts a spectral reflectance of a printed matter printed using an ink amount set from the ink amount set that is a combination of ink amounts of inks used for printing, includes, in an ink amount space in which a plurality of lattice points are disposed, selecting an ink amount set from an ink amount set associated with each lattice point, according to a predetermined selection rule, acquiring a spectral reflectance of a color chart printed on a printing medium using the selected ink amount set, and setting the teacher data using the selected ink amount set as an input value and the acquired spectral reflectance as an output value.