Face Region Detection With Local Reshaping for Stable Video Enhancement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection methods struggle with generalizing skin tone variations, computational expense in video segmentation, and neural networks causing flickering artifacts due to missing detections and temporal inconsistency.
Innovation Solution
A method for face region detection using histogram analysis and local reshaping, adjusting contrast and saturation separately for faces, based on bounding boxes and confidence levels, with adaptive sorting and probability propagation in YUV color space to enhance temporal stability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep convolutional neural networks are used for face region detection, then detection accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent segments the image processing task by first detecting face bounding boxes using a simplified method, then applying histogram analysis only within those bounded regions. This divides the complex full-image analysis into smaller, more manageable face-specific analysis, reducing overall computational complexity while maintaining detection accuracy.
Solution Approach 2:
The patent applies different processing strategies to different regions: face regions receive specialized histogram-based analysis with targeted contrast and saturation adjustments, while non-face regions use standard processing. This local differentiation improves efficiency by avoiding unnecessary complex computations in non-face areas.
2Manufacturing precision
If global contrast enhancement is applied to the entire image, then overall image quality is improved, but face details such as wrinkles and spots become over-enhanced
Solution Approach 1:
The patent implements local quality control by calculating separate histograms for face regions and applying contrast enhancement only to those regions. The contrast adjustment factor is computed based on face-specific histogram statistics, ensuring that face details are enhanced appropriately without the excessive enhancement that would occur with global processing.
Solution Approach 2:
The patent segments the image into face and non-face regions using bounding boxes and probability maps. This segmentation allows independent processing of face regions with specialized contrast adjustment, preventing the over-enhancement of facial defects that would result from uniform global enhancement while still improving overall image quality.
3Productivity
If skin tone detection is used for face region identification, then detection speed is improved, but reliability decreases due to variability in skin tone across different individuals and lighting conditions
Solution Approach 1:
The patent changes from fixed skin tone parameter detection to dynamic histogram-based parameter analysis. By computing histograms for each face region and using these to determine contrast adjustment factors and detection thresholds, the system adapts to varying skin tones and lighting conditions, improving reliability while maintaining the speed benefits of bounding box-based initial detection.
Data Source
AI summary
Methods and corresponding systems to process face regions are disclosed. The described methods include providing face bounding boxes and confidence levels for the faces, generating a histogram of the pixels and the faces, generating a probability of face, and generating a face probability map. A face contrast adjustment and a face saturation adjustment can be applied to the face probability map.