Spectral Reflectance Estimation for Accurate Color Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital image devices face challenges in converting input colors into device-independent color signals due to the ill-posed problem of input data exceeding output data, resulting in large errors in estimating surface spectral reflectance, making it difficult to achieve accurate color conversion suitable for general electric appliances.
Innovation Solution
A color converting system and method that estimates spectral reflectance in multiple wavelength ranges, synthesizes these reflectances, and applies them to convert input images into device-independent color models, using a matrix creation process to minimize errors by dividing visible light regions and applying kernels to calculate spectral reflectance, with error threshold adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If surface spectral reflectance is estimated based on input color signal using conventional methods, then color conversion to device-independent color is achieved, but large errors occur in surface spectral reflectance estimation due to the ill-posed problem where input data order exceeds output data order
Solution Approach 1:
The patent divides the visible light spectrum into multiple wavelength ranges (e.g., 380-480nm, 480-580nm, 580-780nm) and estimates spectral reflectance separately for each range. This segmentation transforms the ill-posed global estimation problem into multiple better-posed local estimation problems, reducing estimation errors in each segment while maintaining overall spectral accuracy.
Solution Approach 2:
The patent introduces regularization parameters and constraint conditions as additional dimensions to the estimation problem. By adding these dimensional constraints (such as non-negativity constraints and smoothness constraints), the system transforms the underdetermined ill-posed problem into a well-posed optimization problem with unique solutions, thereby improving reliability.
2Measurement precision
If devices for measuring surface spectral reflectance are used, then accurate spectral reflectance data is obtained, but the devices are expensive and not suitable for general electric appliances
Solution Approach 1:
The patent creates a computational model that copies the functionality of expensive spectral reflectance measurement devices. By developing algorithms that estimate spectral reflectance from standard color signals (RGB or CMYK), the system replicates the measurement capability of specialized devices without requiring their physical hardware, making the technology accessible for general electric appliances.
Solution Approach 2:
The patent replaces the mechanical/optical measurement system (spectral reflectance meters) with a computational/mathematical system. Instead of using physical devices to measure spectral reflectance, the invention uses mathematical algorithms processing color signal data to calculate spectral reflectance, thereby eliminating the need for complex measurement hardware.
3Measurement precision
If the visible light region is divided into multiple wavelength ranges and kernels are applied to calculate spectral reflectance, then estimation accuracy is improved, but the processing complexity and computational load increase
Solution Approach 1:
The patent segments the visible spectrum into discrete wavelength ranges and applies kernel functions to each segment. This segmentation strategy improves accuracy by capturing spectral variations in different regions while managing complexity through systematic processing of each segment using standardized kernel-based methods.
Solution Approach 2:
The patent changes the parameter representation from direct spectral measurements to kernel-based functional representations. By using kernel functions with adjustable parameters to model spectral reflectance curves, the system achieves high accuracy while reducing the number of independent parameters that need to be estimated, thereby managing computational complexity.
Data Source
AI summary
A system, medium, and method converting image colors, more particularly, a system, medium, and method converting image colors by estimating the surface spectral reflectance of a subject in one or more wavelength ranges based on an input color signal and by converting the input color signal into a device-independent color signal based on the estimated surface spectral reflectance. The system may include an image input unit receiving an image, a reflectance estimating unit estimating spectral reflectances of the image in one or more wavelength ranges, a reflectance synthesizing unit synthesizing the estimated spectral reflectances into one spectral reflectance, and a color converting unit converting the color of the image on the basis of the synthesized spectral reflectance.