Logarithmic White Balance Correction for Image Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for white balance correction inaccurately determine color temperature due to the complexity of calculating pixel distances from the Plankian locus, especially when the object's color is misinterpreted as light source effects, and the conversion to CIE color space is computationally intensive.
Innovation Solution
The method involves converting the Plankian locus into a horizontal line through logarithmic transformation, allowing for simpler determination of pixel ranges and color temperature by generating logarithm values based on original gray levels and adjusting parameters specific to the camera's optical sensor, which are pre-stored before factory shipment, enabling direct comparison of logarithm values to determine color temperature and perform white balance correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the CIE color space conversion method is used to determine color temperature, then the measurement precision of color temperature is improved, but the device complexity and calculation time increase significantly
Solution Approach 1:
The patent extracts only the necessary information (logarithm values of R, G, B components) from the full CIE color space conversion process. Instead of converting all pixels to CIE space, it uses a simplified approach that extracts log(R), log(G), log(B) values and compares them directly against pre-stored logarithm ranges, significantly reducing computational complexity while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-storing the logarithm values of R, G, B components for different color temperatures during camera manufacturing. This allows the actual image processing to simply compare calculated logarithm values against these pre-stored ranges, eliminating the need for complex real-time CIE conversions and distance calculations.
2Measurement precision
If the Plankian locus distance calculation method is used, then the color temperature determination is more accurate, but the calculation time and processing complexity increase
Solution Approach 1:
The patent extracts the essential comparison task from the complex Plankian locus distance calculation. Instead of calculating distances from the curved Plankian locus for each pixel, it extracts the comparison to pre-stored logarithm ranges, reducing the operation to simple range checking that is much faster computationally.
Solution Approach 2:
The patent performs preliminary action by pre-calculating and storing the logarithm ranges corresponding to different color temperatures during manufacturing. This transforms the runtime task from complex distance calculations to simple range comparisons, significantly reducing processing time.
3Device complexity
If the gray world assumption method is used, then the white balance correction is simpler to implement, but the measurement precision deteriorates when object color affects the image
Solution Approach 1:
The patent applies local quality by considering the specific characteristics of each pixel's R, G, B values individually through logarithm transformation, rather than relying on the global average assumption of the gray world method. This allows the algorithm to account for local color variations while maintaining simplicity.
Solution Approach 2:
The patent changes the parameter representation from raw R, G, B values to their logarithm values. This parameter transformation allows for a simplified comparison mechanism that works independently for each color channel, maintaining algorithm simplicity while improving accuracy by avoiding the pitfalls of the gray world assumption.
Data Source
AI summary
An image processing method is provided. A digital image is outputted by an optical sensor and includes M pixels. The ith pixel among the M pixels has an ith set of original gray levels, wherein i is an integer index ranging from 1 to M. In the method according to this invention, M sets of logarithm values are first generated. The ith set of logarithm values among the M sets of logarithm values is generated based on the ith set of original gray levels and a set of adjusting parameters relative to the optical sensor. Statistics of the M sets of logarithm values is then gathered. According to the statistics result, a target color temperature of the digital image is determined.