Adaptive Face Detection Using Skin Maps and Dynamic Window Increments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection methods in digital image acquisition devices suffer from slow execution times when detecting small faces, making them unsuitable for real-time embedded implementations due to the increased number of sliding windows required.
Innovation Solution
An adaptive method that calculates x and y increments for a detection window based on confidence levels, reducing the number of windows analyzed while maintaining face detection quality, and employing a skin map to prioritize potential face regions, allowing for faster face searching and detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the detection window size decreases to detect small faces, then the detection precision is improved, but the number of sliding windows increases quadratically resulting in slower execution time
Solution Approach 1:
The patent applies skin detection as a preliminary action before face detection. By first identifying skin regions in the image and creating a skin map, the system pre-processes the image to highlight areas likely to contain faces. This preliminary action reduces the search space for subsequent face detection, allowing the system to maintain high detection precision for small faces while avoiding the quadratic increase in processing time that would result from analyzing every possible window position.
2Ease of manufacture
If linear searching with constant increments is used, then the implementation simplicity is improved, but the execution time increases for small faces due to quadratic increase in window count
Solution Approach 1:
The patent introduces dynamic adaptation to the face detection process by using skin detection results to guide window placement and increment selection. Instead of using fixed constant increments throughout the image, the system dynamically adjusts detection parameters based on skin map information. This allows the detector to concentrate computational resources on regions with high probability of containing faces, significantly reducing detection time while maintaining implementation feasibility through the use of adaptive rather than purely static parameters.
Data Source
AI summary
A method of detecting a face in an image includes performing face detection within a first window of the image at a first location. A confidence level is obtained from the face detection indicating a probability of the image including a face at or in the vicinity of the first location. Face detection is then performed within a second window at a second location, wherein the second location is determined based on the confidence level.