Dynamic Video Equalization Using Face-Tracking Parameter Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video conferencing images often suffer from inferior quality due to inadequate allocation of image parameters, with the face of participants being of lower quality compared to the background, exacerbated by limited data transmission and poor lighting conditions.
Innovation Solution
Implementing a system that uses face tracking technology to dynamically allocate a greater range of image parameters such as colors, brightness levels, and resolution to the face area, rather than equally distributing them across the entire image, thereby optimizing the image quality of the face.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If face tracking technology is used to center and zoom into the face, then the face dominance in the image is improved, but the image information for background objects is reduced and the overall image quality is compromised
Solution Approach 1:
The patent divides the image into two distinct segments: the face region and the background region. By segmenting the image, the system can apply different image parameter allocations to each segment, allowing high-quality representation of the face while maintaining acceptable background representation. This segmentation enables selective optimization without sacrificing one region for the other.
Solution Approach 2:
The patent applies local quality by allocating different ranges of image parameters to different regions. Specifically, a greater range of image parameters is allocated to the face region compared to the background region. This allows the face to receive enhanced detail and quality where it is most important, while the background receives sufficient but reduced quality parameters.
2Reliability
If equal allocation of image parameters is used across the entire image, then the overall image quality is maintained, but the face quality is compromised due to limited data transmission bandwidth
Solution Approach 1:
The patent implements local quality by differentiating parameter allocation between regions. The face region receives a greater allocation of image parameters (higher quality) while the background receives a reduced allocation. This unequal distribution optimizes the limited bandwidth by concentrating resources where they provide the most value - in the face region.
Solution Approach 2:
The patent applies partial action by focusing image parameter allocation only where necessary. Instead of uniformly distributing all parameters across the entire image, the system allocates excessive (relative to needs) parameter range to the face region to ensure high quality, while using minimal sufficient parameters for the background.
3Reliability
If backlight compensation is applied to equalize the overall image histogram, then the overall image quality is improved, but the face quality remains relatively low due to peak intensity removal
Solution Approach 1:
The patent segments the image processing into two stages: first applying backlight compensation to the entire image to improve overall quality, then applying selective optimization to the face region identified by face tracking. This segmentation allows the benefits of global equalization to be retained while adding local enhancement to the face.
Solution Approach 2:
The patent performs preliminary backlight compensation on the entire image before applying face-specific optimization. This preliminary action ensures that the overall image histogram is equalized and peak intensities are removed, creating a foundation of improved overall quality upon which the face-specific enhancement is built.
Data Source
AI summary
The present invention provides for the dynamic video equalization of images. Face tracking is used to identify a portion of an image corresponding to a human face. Those areas of the image identified as corresponding to a human face are optimized as compared to other areas of the image. Optimization is performed by allocating a greater number of image parameters to the area of the image corresponding to a human face than are allocated to other areas of the image. Accordingly, the portion of an image containing the human face is of higher quality as compared to other portions of the image.