Dynamic White Balance via Illuminant Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automatic white balancing (AWB) methods assume a single illuminant source, leading to less accurate estimations and unnatural image coloring, especially in images captured by smartphone cameras, as they fail to account for combinations of different illuminant sources.
Innovation Solution
The method involves a color space diagram with pre-defined white areas for different illuminant sources, calculating weighted mean RGB values of pixels in these areas to determine the illuminant sources, and applying correction vectors to adjust the white pixel position based on brightness and color temperature, allowing for the possibility of multiple illuminant sources and improving image balance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional AWB methods assume a single illuminant source, then the algorithm complexity is reduced, but the white balance accuracy deteriorates
Solution Approach 1:
The patent segments the image into multiple regions, each potentially illuminated by different illuminant sources. Instead of treating the entire image as a single illuminant type, the algorithm divides it into region 1, region 2, etc., where each region can have its own illuminant characteristics. This segmentation allows accurate white balance for each region while managing complexity through localized processing.
Solution Approach 2:
The patent applies local quality by allowing different parts of the image (different regions) to have different white balance characteristics. Each region is processed with its own illuminant estimation and correction factors, rather than applying a uniform white balance setting to the entire image. This ensures that local color accuracy is maintained in areas with different lighting conditions.
2Productivity
If conventional AWB methods use a single illuminant assumption, then the processing speed is improved, but the image color naturalness deteriorates
Solution Approach 1:
The image is divided into multiple regions that can be processed independently with different illuminant assumptions. This segmentation enables parallel processing of regions, maintaining processing speed while improving color naturalness by allowing each region to be corrected according to its specific lighting conditions.
Solution Approach 2:
The patent applies white balance correction selectively to different regions rather than uniformly to the entire image. By applying correction only where needed and with region-specific parameters, the algorithm maintains efficiency while achieving more natural colors in areas with mixed illuminant sources.
3Power
If the AWB algorithm processes the whole image uniformly, then the computational load is reduced, but the accuracy in mixed lighting conditions deteriorates
Solution Approach 1:
The patent divides the image into multiple regions and processes each region separately to identify its specific illuminant source. This segmentation approach reduces the computational load compared to analyzing the entire image uniformly, while simultaneously improving accuracy by allowing each region to be analyzed according to its local lighting characteristics.
Solution Approach 2:
The algorithm applies local quality by estimating illuminant sources and applying corrections specific to each region rather than using a uniform approach for the entire image. This ensures that regions with different lighting conditions are processed with appropriate local parameters, improving overall accuracy while managing computational resources efficiently.
Data Source
AI summary
A method of white balancing an image includes mapping pixels of the image to a color space diagram. Each of the pixels of the image include a red (“R”), a green (“G”), and a blue (“B”) subvalue. A first central tendency of each of the RGB subvalues of pixels mapped in a first pre-defined region of the color space diagram is determined and a second central tendency of each of the RGB subvalues of pixels mapped in a second pre-defined region of the color space diagram is determined. The first pre-defined region is associated with a first illuminating source and the second pre-defined region is associated with a second illuminating source. RGB values of a white pixel are generated based on the first and second central tendencies.