Nonlinear Color Transformation for White Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional white balancing techniques fail to accurately correct the color balance of images that have undergone nonlinear color transformations, leading to incorrectly white-balanced output images, as they apply linear transformations to images that have already been nonlinearly manipulated.
Innovation Solution
The development of an image editing system that computes a nonlinear color transformation for input images based on training images with similar color characteristics, using a correction matrix and kernel function to apply a nonlinear transformation, enabling accurate white balancing even for images subjected to previous nonlinear color manipulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional linear color transformation is applied for white balancing, then the process is simple and fast, but the white balancing accuracy deteriorates for images that have undergone nonlinear color transformations
Solution Approach 1:
The patent changes the mathematical parameters of color transformation from linear to nonlinear by introducing a kernel function with polynomial terms (e.g., R^2, G^2, B^2, RG, GB, BR) and adaptive weighting coefficients. This allows the system to model complex color relationships and accurately reverse nonlinear color transformations that conventional linear methods cannot handle.
Solution Approach 2:
The patent introduces a training dataset as an intermediary between the input image and the final color transformation. The training dataset contains pre-computed color transformations for images with known color characteristics, which serve as a reference to guide the computation of appropriate transformation parameters for new images, improving accuracy without requiring complex real-time analysis.
2Measurement precision
If in-camera white balancing is used, then the process is automated and fast, but the accuracy deteriorates because the camera cannot distinguish between white background under yellow light and yellow background under white light
Solution Approach 1:
The patent implements feedback by using the training dataset to continuously refine and improve white balancing accuracy. The system learns from previously processed images with known color characteristics and uses this information to guide subsequent white balancing operations, progressively improving accuracy without requiring complex manual intervention.
Solution Approach 2:
The patent creates a virtual copy of the color transformation process through the training dataset, which stores pre-computed transformations for various illumination conditions and color characteristics. This allows the system to reference and replicate accurate transformations for new images without requiring real-time analysis of complex color relationships.
3Measurement precision
If post-capture white balancing is applied to images with nonlinear color transformations, then the white balancing step can be added, but the overall process complexity increases and mathematical correctness is compromised by applying linear transformation to nonlinear data
Solution Approach 1:
The patent fundamentally changes the mathematical parameters from linear to nonlinear transformations by incorporating polynomial kernel functions with multiple interaction terms. This allows the system to correctly model and reverse the nonlinear color transformations that occur during camera processing, maintaining mathematical correctness throughout the pipeline.
Solution Approach 2:
The patent performs preliminary computation of color transformations during the training phase, where pre-computed transformations are stored in the training dataset. This allows the system to prepare and store accurate transformation parameters in advance, which can then be efficiently applied to new images without requiring complex real-time computation, thus reducing operational complexity.
Data Source
AI summary
Techniques are described for white balancing an input image by determining a color transformation for the input image based on color transformations that have been computed for training images whose color characteristics are similar to those of the input image. Techniques are also described for generating a training dataset comprising color information for a plurality of training images and color transformation information for the plurality of training images. The color information in the training dataset is searched to identify a subset of training images that are most similar in color to the input image. The color transformation for the input image is then computed by combining color transformation information for the identified training images. The contribution of the color transformation information for any given training image to the combination can be weighted based on the degree of color similarity between the input image and the training image.