Image Processing Apparatus White Balance Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques face challenges in accurately performing white balance correction, especially in scenarios with multiple light sources, such as outdoors, where small achromatic regions make it difficult to estimate the light source and reproduce original colors.
Innovation Solution
An image processing apparatus that includes a detection unit to identify specific areas in a captured image and a determination unit using a learning model trained by machine learning to determine the light source color, enabling accurate white balance correction by estimating the light source based on characteristic colors from detected areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If white balance processing is performed based on achromatic color regions, then accurate light source estimation is achieved, but the method fails when the achromatic region is extremely small
Solution Approach 1:
The patent changes the parameter used for light source estimation from achromatic color regions to skin color regions. By detecting skin color areas and using their color characteristics, the system can estimate the light source even when traditional white balance reference areas are unavailable or extremely small.
Solution Approach 2:
The patent uses skin color areas, which are always present in portrait photography, as a substitute for traditional white balance references. This approach leverages readily available image content (skin tones) rather than requiring specific dedicated reference areas.
2Ease of operation
If skin color area is used for white balance processing, then processing can be performed without sufficient achromatic regions, but the conversion is not based on light source determination leading to inadequate color reproduction in complex lighting
Solution Approach 1:
The patent implements a feedback mechanism where the detected skin color information is used to estimate the light source characteristics, which then guides the white balance adjustment. The system continuously refines the light source estimation based on the color characteristics observed in the skin regions.
Solution Approach 2:
The system changes from direct skin color conversion to light source-based estimation. By first determining the light source type and characteristics from skin color regions, and then applying appropriate white balance corrections based on that light source identification, the system achieves both operational feasibility and color accuracy.
3Device complexity
If traditional white balance methods are used in environments with multiple light sources, then simple processing is applied, but the original colors cannot be reproduced due to inability to identify dominant light source
Solution Approach 1:
The patent changes the approach from assuming uniform lighting to analyzing color distribution patterns across skin regions. By examining variations in skin tone colors and using machine learning to identify the dominant light source characteristics, the system can handle complex multi-light-source environments.
Solution Approach 2:
The patent adds a new dimension of analysis by using machine learning models to interpret skin color data in terms of light source characteristics. This transforms the problem from simple color averaging to sophisticated pattern recognition that can distinguish between multiple light sources.
Data Source
AI summary
An image processing apparatus includes a detection unit configured to detect a specific area from a captured image, and a determination unit configured to determine, based on the detected specific area, a light source color including a characteristic color by using a learning unit trained in advance by machine learning.