White Balance Adjustment Using Face Silhouette Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video conferencing in home environments with uneven lighting conditions leads to poor image quality due to light hotspots and unbalanced lighting, as conventional white balance solutions are designed for evenly distributed office lighting.
Innovation Solution
A system that analyzes a limited portion of the image, such as the silhouette of a human face, to determine a white balance correction and applies it across the image, while de-emphasizing areas outside the analyzed perimeter through dimming or blurring to improve the quality of the video conference image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional white balance solutions are used that analyze all light in the field of view, then the system can process the entire image, but the image quality deteriorates due to light hotspots and unbalanced lighting in home environments
Solution Approach 1:
The patent segments the field of view into multiple zones, identifying a primary zone containing the user's face and secondary zones containing background elements. White balance analysis is performed separately on the primary zone, isolating it from problematic lighting in secondary zones. This segmentation allows the system to compute white balance based only on relevant areas, eliminating the harmful impact of light hotspots in the background.
Solution Approach 2:
The patent applies different processing quality levels to different regions of the image. The primary zone containing the user's face receives full white balance correction and high-quality processing, while secondary zones with problematic lighting are either excluded from analysis or given reduced processing priority. This local quality approach ensures optimal image quality for the most important subject while managing the limitations imposed by uneven lighting.
2Adaptability or versatility
If white balance analysis includes the entire field of view, then all areas are processed uniformly, but the user's face quality deteriorates when background lighting is unbalanced
Solution Approach 1:
The system segments the image to identify the user's face as a distinct region of interest, separating it from the background. This allows the white balance algorithm to adapt to the specific lighting conditions affecting the face while ignoring or separately processing background areas with different lighting characteristics. The segmentation enables the system to maintain high precision for face imaging even when overall scene lighting is unbalanced.
Solution Approach 2:
The patent applies enhanced processing quality specifically to the user's face region while using standard or reduced processing for other areas. The white balance correction is locally optimized for the face zone, ensuring high manufacturing precision for the most important subject. This local quality approach allows the system to adapt to various lighting conditions while maintaining consistent high quality for the user's image.
3Area of stationary object
If the system processes the entire image for white balance correction, then comprehensive coverage is achieved, but processing complexity increases and quality improves only marginally in problematic areas
Solution Approach 1:
The patent divides the image processing task into segments: first identifying the primary zone containing the user's face, then performing white balance analysis only on that segment. This segmentation dramatically reduces the amount of data that needs to be processed while still achieving comprehensive coverage of the important subject. The system maintains full image coverage for display purposes but applies complex processing only where necessary.
Solution Approach 2:
The system performs partial white balance correction, applying full processing only to the primary zone containing the user's face rather than attempting to uniformly correct the entire field of view. This partial action approach achieves the essential goal of improving user image quality without the excessive complexity of processing every pixel in the scene, particularly in areas where lighting conditions are beyond the system's control.
Data Source
AI summary
Videoconferencing images captured by a video camera in suboptimal lighting conditions are enhanced by analyzing light associated with a portion of the video camera field of view and applying that analysis for white balance of the entire field of view. For example, the face or silhouette of a videoconference participant is detected within the camera field of view so that light within a perimeter of the face or silhouette is analyzed to determine a white balance correction for application to the entire field of view. To reduce the impact of the white balance correction factor based upon the area within a silhouette participant, the image outside the perimeter may be subdued, such as with blurring or dimming.