Image Processing Device White Balance Gain Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately determining the type of light source in scenes with objects of similar color distributions, such as skin tones and tungsten light, which hinders automatic white balance (AWB) correction.
Innovation Solution
An image processing device acquires and processes imaging signals from flash emission and non-emission images to isolate the influence of flash light, allowing for the calculation of white balance gains specific to each area of an image, thereby improving AWB correction by distinguishing object colors from illumination effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object color component data is acquired using flash emission and non-emission images, then the accuracy of light source estimation is improved, but the device complexity increases
Solution Approach 1:
The image processing is divided into distinct segments: acquiring flash emission image data, acquiring flash non-emission image data, calculating difference image data, and generating illumination component data. This segmentation allows each processing stage to be optimized independently, improving light source estimation accuracy while managing complexity through modular processing steps.
Solution Approach 2:
The patent performs preliminary processing by calculating difference image data between flash emission and non-emission images before final white balance correction. This preliminary action isolates the flash light influence, enabling more accurate light source estimation and subsequent AWB correction based on object color component data.
2Reliability
If white balance correction is performed using conventional methods, then the processing speed is maintained, but the reliability of AWB correction deteriorates in scenes with similar color distributions
Solution Approach 1:
The patent introduces difference image data as an intermediary between the raw flash emission/non-emission images and the final white balance correction. This intermediary representation isolates the illumination component, serving as a mediator that enables reliable light source estimation even in challenging scenes with similar color distributions like skin tones and tungsten light.
Solution Approach 2:
The patent changes the parameter representation by transforming raw image data into object color component data through mathematical operations (difference calculation, multiplication by spectral data). This parameter transformation enables more reliable AWB correction by separating object color from illumination effects, allowing accurate white balance gains to be calculated.
3Measurement precision
If flash light influence is isolated using difference image data, then the object color estimation accuracy is improved, but the loss of information increases
Solution Approach 1:
The patent selectively discards the flash light influence component from the image data while recovering and preserving the object color information. By calculating difference image data and multiplying by flash spectral data, the method discards unwanted illumination effects while recovering pure object color component data for accurate white balance correction.
Data Source
AI summary
An image processing device includes a first image acquisition unit that acquires each of a first imaging signal indicating a flash emission image and a second imaging signal indicating a flash non-emission image, a second image acquisition unit that acquires a third imaging signal indicating a difference between the first imaging signal and the second imaging signal that have been acquired, a third image acquisition unit that acquires a fourth imaging signal obtained by multiplying the acquired third imaging signal by a white balance gain for a flash for removing an influence due to color of the flash light, a color signal acquisition unit that acquires a color signal indicating a first color of each area in an imaging screen on the basis of the acquired fourth imaging signal, and a first white balance gain calculation unit.