Hybrid Color Mapping for Accurate Image Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image color correction techniques face challenges in achieving accurate and flexible color correction, with global methods being less accurate and prone to color artifacts, while point-to-point methods require pixel correspondence and are not always feasible.
Innovation Solution
A method that combines global color mapping with point-to-point color mapping, using a pseudo-inverse technique to compute a color mapping function that minimizes error between the globally-mapped and input images, ensuring accurate color correction without requiring pixel correspondence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If point-to-point color mapping methods are used, then color correction accuracy is improved, but pixel correspondence is required which reduces flexibility and ease of operation
Solution Approach 1:
The patent segments the color mapping process into two independent stages: first performing global color mapping to correct overall color statistics, then applying point-to-point color mapping only to specific color values. This segmentation allows the system to achieve high accuracy where needed while maintaining flexibility and automation capability where pixel correspondence is not required.
2Ease of operation
If global color mapping methods are used, then flexibility and ease of operation are improved, but color correction accuracy deteriorates and color artifacts are introduced
Solution Approach 1:
The patent merges global color mapping and point-to-point color mapping into a hybrid approach. Global color mapping is used to establish overall color statistics and handle cases where pixel correspondence is unavailable, while point-to-point color mapping is applied to refine specific color value mappings. This combination achieves both flexibility and high accuracy while reducing color artifacts.
3Adaptability or versatility
If Histogram Specification is used for global color mapping, then versatility is improved, but color artifacts are introduced and automation becomes difficult
Solution Approach 1:
The patent converts the potential harm of color artifacts from Histogram Specification into a benefit by using the globally-mapped image as training data for subsequent point-to-point color mapping. The point-to-point stage refines the color values and removes artifacts while maintaining the versatility and automation capabilities established by the global mapping stage.
Data Source
AI summary
A reference value for a color statistic and an input image sample are received. A global color mapping is performed between the input image sample and the reference value to obtain a globally-mapped input image sample, the color statistic of the globally-mapped input image sample substantially matching the reference value and each pixel location in the globally-mapped input image sample having a correspondence with a corresponding pixel location in the input image sample. A point-to-point color mapping function minimizing for all pixels an error between the globally-mapped input image sample and the input image sample mapped according to the color mapping function is computed, the color mapping function transforming a color value of each pixel location in the input image sample to substantially match a color value of the corresponding pixel location in the globally-mapped input image sample.