Face Mask Detection for Accurate Exposure and White Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face masks disrupt automatic exposure and white balance adjustments in image processing systems, causing brightness and color abnormalities in images of individuals wearing masks.
Innovation Solution
Detecting a face mask by comparing hue, saturation, and brightness differences between the upper and lower halves of a face, and adjusting automatic exposure and white balance only on the uncovered face area to maintain accurate skin color and brightness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If automatic exposure and white balance adjustments are applied to the entire face area, then the overall image brightness and color are optimized, but the skin color and brightness become abnormal due to the face mask's different color and brightness characteristics
Solution Approach 1:
The face area is segmented into two regions: the upper area (uncovered face) and the lower area (face mask). Different image processing operations are applied to each region independently. The upper area undergoes automatic exposure and white balance adjustments, while the lower area is excluded from these operations, preventing the mask's color and brightness characteristics from affecting the skin tone accuracy.
Solution Approach 2:
Different processing quality and parameters are applied to different parts of the image. The upper face area receives full automatic exposure and white balance processing to maintain accurate skin color, while the lower face area (mask) is processed differently or not at all, allowing each region to have optimized local characteristics without compromising the other.
2Manufacturing precision
If deep learning-based methods are used for face mask detection and image processing, then accurate skin color and brightness can be maintained, but power consumption increases significantly
Solution Approach 1:
The patent uses simpler, less computationally intensive methods compared to deep learning models. By employing traditional image processing techniques and basic color space conversions, the system achieves accurate skin color maintenance with significantly lower power consumption, replacing complex and energy-intensive deep learning approaches with more efficient alternatives.
3Productivity
If the entire face area is processed for automatic exposure and white balance, then processing speed is maintained, but brightness and color abnormalities occur due to the face mask
Solution Approach 1:
The processing area is segmented to exclude the face mask region from automatic exposure and white balance operations. By identifying and isolating the mask area, the system processes only the appropriate regions (upper face) for these operations, maintaining processing efficiency while eliminating the harmful brightness and color abnormalities that would result from processing the entire face area uniformly.
Data Source
AI summary
Methods, apparatus, systems, and articles of manufacture are disclosed to automatically process an image based on a detection of a face mask. An example article of manufacture include instructions that, when executed, cause programmable circuitry to at least: map a characteristic of an upper area of a face of a person in an image to a color plot including a face skin tone cone; map a characteristic of a lower area of the face to the color plot; and identify a presence or an absence of a face mask based on the respective positions of the characteristic of the upper area and the characteristic of the lower area relative to the skin tone cone.