Real-time Face Tracking via Adaptive Subsampling and Skin Tone Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection algorithms, such as Viola-Jones, require significant computational resources and result in high false positive rates when implemented in resource-restricted embedded systems like digital cameras, making real-time face tracking impractical.
Innovation Solution
The method involves subsampling images to reduce processing load, applying face detection only to regions with skin tones, and using a combination of fixed-size and variable-size classifiers to efficiently detect and track faces across different scales, while adjusting skin segmentation and contrast enhancement based on lighting conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If Viola-Jones face detection algorithm is used, then detection speed is improved, but computational resource consumption increases significantly
Solution Approach 1:
The patent divides the image into multiple regions and applies face detection only to selected regions rather than the entire image. This segmentation approach reduces the computational domain while maintaining detection capability, directly addressing the contradiction between detection speed and resource consumption by processing only relevant portions of the image.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image based on their likelihood of containing faces. High-priority regions receive full Viola-Jones processing while low-priority regions receive reduced or no processing. This local quality differentiation allows the system to maintain high detection speed in critical areas while conserving computational resources in less important areas.
2Measurement precision
If full image processing is applied, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the image into multiple regions and processes only selected segments for face detection. This reduces the total processing time while maintaining accuracy in the processed segments, resolving the contradiction between full-image accuracy and processing time by focusing computational effort on regions most likely to contain faces.
Solution Approach 2:
The patent applies partial processing to the image by selecting only certain regions for full Viola-Jones analysis rather than processing the entire image. This partial action approach achieves sufficient detection accuracy for practical purposes while dramatically reducing processing time, accepting that some regions may not be fully processed.
3Reliability
If skin tone region filtering is applied, then false positive rate is reduced, but detection coverage may be limited
Solution Approach 1:
The patent applies skin tone filtering selectively to specific regions of the image rather than uniformly across the entire image. This allows the system to reduce false positives in regions where skin tone analysis is reliable while maintaining broader detection coverage in regions where skin tone filtering might be less effective, thus balancing reliability and adaptability.
Data Source
AI summary
An image processing apparatus for tracking faces in an image stream iteratively receives an acquired image from the image stream including one or more face regions. The acquired image is sub-sampled at a specified resolution to provide a sub-sampled image. An integral image is then calculated for a least a portion of the sub-sampled image. Fixed size face detection is applied to at least a portion of the integral image to provide a set of candidate face regions. Responsive to the set of candidate face regions produced and any previously detected candidate face regions, the resolution is adjusted for sub-sampling a subsequent acquired image.