Automatic White Balancing Gray Region Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic white balancing techniques face challenges in accurately identifying gray pixels due to the inclusion of non-gray colors within the quadrilateral region used to enclose white points, leading to incorrect white balancing results under different light sources.
Innovation Solution
The method involves defining a gray region by comparing blue/green and red/green ratios to a plurality of white points, using color distance metrics and a threshold value to accurately identify gray pixels, which are then input to an automatic white balance algorithm for precise white balancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a large quadrilateral region is used to enclose white points in B-G versus R-G space, then all white points for different light sources are enclosed, but non-gray colors are also erroneously included leading to incorrect white balancing
Solution Approach 1:
The patent divides the gray region identification into multiple steps: first identifying a preliminary gray region using a quadrilateral, then further segmenting it by calculating color distance metrics for each pixel to white points, and finally filtering pixels based on threshold comparisons. This segmentation process separates true gray pixels from non-gray colors that were erroneously included in the initial quadrilateral region.
Solution Approach 2:
The patent applies different evaluation criteria to different regions within the quadrilateral. Instead of treating all pixels in the quadrilateral equally, it calculates color distance metrics for each pixel to determine its local properties. Pixels closer to white points are identified as gray, while those farther away are excluded, creating a refined gray region with locally optimized quality.
2Measurement precision
If a small quadrilateral region is used to enclose white points, then only pixels close to white points are included, but some white points for different light sources may not be enclosed
Solution Approach 1:
The patent performs a preliminary action by first creating a quadrilateral that encloses all expected white points for different light sources. This preliminary region ensures comprehensive coverage, and then subsequent refinement steps remove erroneous inclusions. The preliminary quadrilateral acts as a safe initial boundary that guarantees no white points are missed.
Solution Approach 2:
The patent uses feedback by calculating color distance metrics for each pixel in the preliminary gray region and comparing these distances to threshold values. This feedback mechanism allows the algorithm to identify which pixels truly represent gray colors and which do not, adjusting the final gray region based on quantitative measurements of color proximity to white points.
Data Source
AI summary
Examples of the disclosure accurately define a gray region for automatic white balancing. Ratios corresponding to possible gray areas are white balanced before color distance metrics are calculated with respect to white points under different light sources. The color distance metrics of each point in the image data are compared with a predetermined threshold value to identify a gray region. Gray points within a captured image are identified using the gray region and input to any AWB method to accurately white balance the captured image.