Auto White Balance Control for Single-Tone Image Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing auto white balance control algorithms in digital image processing are prone to errors due to the influence of single-tone images, which can lead to defective white balance and increased system load, especially when determining transitions from composite to single-tone images.
Innovation Solution
An apparatus and method that generate histograms of Cb and Cr values to determine if an image is single-tone, excluding pixels with high frequencies and small ranges to prevent erroneous white balance, and computing auto white balance gains using previous frame data if the image is deemed single-tone or if excessive pixels are excluded.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If adjacent pixel comparison method is used to recognize single-tone image, then single-tone image recognition accuracy is improved, but system efficiency deteriorates due to continuous error accumulation and high computational load
Solution Approach 1:
The patent extracts the essential characteristic of single-tone images by generating histograms of color difference signals (Cb and Cr values) and identifying peak frequencies. Instead of comparing all adjacent pixels, the method extracts only the dominant color tone information from the histogram peaks, significantly reducing computational complexity while maintaining recognition accuracy.
Solution Approach 2:
The patent changes the parameter space from individual pixel comparisons to histogram-based frequency analysis. By transforming the problem from spatial domain (pixel-by-pixel comparison) to frequency domain (histogram peak detection), the method achieves both accuracy and efficiency by analyzing the distribution pattern of color values rather than individual pixel relationships.
2Measurement precision
If outline area determination and block division method is used to recognize single-tone image, then recognition accuracy is improved, but system load increases due to complex processing steps
Solution Approach 1:
The patent removes the complex steps of outline determination and block division by directly analyzing the histogram of the entire image. The method extracts only the necessary information (dominant color tone from histogram peaks) without performing unnecessary segmentation and outline detection, thereby reducing system load while maintaining recognition capability.
Solution Approach 2:
The histogram-based method serves multiple functions simultaneously: it identifies single-tone images, determines the dominant color tone, and provides the basis for white balance adjustment all in one processing step. This universal approach replaces the multi-step process of outline detection, block division, and separate analysis, simplifying the overall system.
3Ease of operation
If mean color value method is used for auto white balance control, then processing simplicity is improved, but white balance accuracy deteriorates when single-tone images are present
Solution Approach 1:
The patent applies local quality by focusing analysis on specific regions of the color distribution (histogram peaks) rather than uniformly processing all pixels. By identifying and analyzing only the dominant color tones represented by histogram peaks, the method achieves accurate white balance adjustment for single-tone images while maintaining simplicity, avoiding the need to process every pixel equally.
Solution Approach 2:
The patent uses partial action by computing white balance gains based only on the dominant color tones identified from histogram peaks, rather than using all color information in the image. This selective approach prevents single-tone regions from excessively influencing the overall white balance calculation, improving accuracy while maintaining processing efficiency.
Data Source
AI summary
An apparatus for auto white balance includes a histogram generation unit writing respective histograms of Cb values and Cr values with regard to pixels included in an input image frame, a single-tone determination unit comparing respective maximum values of the histograms of Cb values and Cr values with a first threshold value, and determining the input image frame as a single-tone image if at least one of the maximum values is higher than the first threshold value, and an auto white balance gain computation unit computing and an auto white balance gain by using color information of an input image frame and outputting the computed auto white balance gain, wherein if the input image frame is determined to be a single-tone image, the auto white balance gain computation unit outputs an auto white balance gain computed with respect to a previous input image frame.