Modified Grey World White Balance Using Geometric Distortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital image white balance correction methods, such as the grey world assumption, fail to accurately balance images with large objects due to color dominance, leading to incorrect settings.
Innovation Solution
A method that splits images into blocks, applies geometric distortion criteria to detect grey blocks, and uses a modified grey world process to generate and refine gain values, removing frequent colors and normalizing hues to improve white balance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the grey world assumption is used for white balance correction, then the process is simple and fast, but the accuracy deteriorates when large objects dominate the image
Solution Approach 1:
The image is divided into multiple blocks, and each block is evaluated independently using geometric distortion criteria to identify grey blocks. This segmentation allows the algorithm to exclude regions dominated by large objects while still processing the overall image efficiently.
Solution Approach 2:
Different regions of the image are treated differently based on their local characteristics. Grey blocks identified by the geometric distortion criteria are selected for white balance calculation, while regions with large objects or false positives are excluded. This local quality approach ensures that only suitable regions contribute to the white balance determination.
2Device complexity
If traditional grey world method is applied, then computational complexity is low, but false positives increase due to color saturation and large objects
Solution Approach 1:
The geometric distortion criteria is applied beforehand to identify and exclude regions with large objects before performing the white balance calculation. This preliminary action prevents false positives from affecting the final result, improving reliability without significantly increasing complexity.
Solution Approach 2:
The algorithm uses feedback from the geometric distortion measurement to iteratively refine the selection of grey blocks. By continuously evaluating which blocks meet the criteria and adjusting the selection accordingly, the system reduces false positives while maintaining computational efficiency.
Data Source
AI summary
A method, apparatus and computer program product for correcting white balance in a digital image are described. The method comprises determining a number of grey blocks in the digital image (s102), each block comprising a subset of pixels of the image, and if the number of grey blocks is less than a predetermined value, scaling one or more color channels of the digital image using a first gain value to generate a first intermediate image (s201A) and scaling one or more color channels of the digital image using a second gain value to generate a second intermediate image (s201B). A number of grey blocks in each of the first and second intermediate images (s202A, s202B) is determined and the gain value resulting in the intermediate image having a greater number of grey blocks is set as the reference gain value (s203). A grey world process is performed on the intermediate image having a greater number of grey blocks (s205). Also described are a method, apparatus and computer program product for computing gain values for correcting a white balance of a digital image. The method comprises receiving an image divided into a number of blocks, removing the three most frequently occurring colors from the image (s204), and scaling one or more color channels of the image by respective reference gain values derived by a grey world process performed on the image.