Automatic White Balance Using Segmented Color Space Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Prior automatic white balance methods in image capture systems are prone to being influenced by strong colors, leading to incorrect gain adjustments and failure to accurately identify the illuminant source, as they either include all pixels in averaging calculations or use a single white area in color space diagrams that can misclassify non-white pixels.
Innovation Solution
The implementation of a color space diagram with predefined white areas for different illuminant sources, where the Green/Red (G/R) and Green/Blue (G/B) ratios of pixels are used to identify white pixels and determine the illuminant source, allowing for selective averaging of white pixel values to calculate necessary gain adjustments for each color channel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If all pixels are included in averaging calculations to determine gain adjustments, then the white balance calculation is simple, but strong color objects skew the average RGB values and cause color accuracy loss
Solution Approach 1:
The patent segments the image pixels into different categories based on their color characteristics. Instead of treating all pixels uniformly, it identifies and separates white pixels from colored pixels using color space analysis. This segmentation allows the averaging calculation to include only relevant white pixels, eliminating the skewing effect of strong color objects while maintaining calculation feasibility.
Solution Approach 2:
The patent applies local quality by treating different regions of the color space differently. It defines specific white areas in the color space diagram where only pixels falling within these regions are considered for averaging. This localized approach ensures that only pixels with appropriate white characteristics contribute to the gain calculation, improving color accuracy without requiring complex processing of all pixels.
2Device complexity
If a single white area is defined in color space diagram for all illuminant types, then the method is simple to implement, but non-white pixels may be misclassified as white pixels leading to incorrect RGB averaging
Solution Approach 1:
The patent implements dynamics by making the white area definition adaptive to different illuminant conditions. Instead of using a fixed single white area, it provides multiple predefined white areas corresponding to different illuminant types (daylight, fluorescent, incandescent, etc.). The system dynamically selects the appropriate white area based on the detected illuminant, allowing accurate white pixel identification across varying lighting conditions without requiring complex real-time adjustments.
3Device complexity
If a single white area template is used for detecting white pixels across all illuminant types, then the detection method is unified and simple, but the system cannot ascertain the illuminant source of the image
Solution Approach 1:
The patent adds another dimension to the white pixel detection process by incorporating illuminant type classification. Instead of merely detecting white pixels, the system determines which predefined white area each pixel falls into, thereby inferring the illuminant source. This dimensional extension transforms a simple detection task into a dual-function system that simultaneously identifies white pixels and ascertains illuminant conditions without significantly increasing complexity.
Data Source
AI summary
A method of automatic white balance for an image capture system is disclosed. The automatic white balance mechanism ascertains the illuminant source of an image by analyzing the number of white pixels within a predefined white area in a color space diagram. The automatic white balance mechanism also determines gain adjustments based on the evaluating the average RGB values to achieve white balance.