Depth-Camera Liveness Detection for Spoof-Resistant Access Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition systems for access control are vulnerable to spoofing techniques, such as using still images or digital displays, and are challenged by noise, poor camera quality, and lighting issues, leading to increased costs for improved performance.
Innovation Solution
A method utilizing a depth camera to generate depth frames, create point clouds, and apply geometric characteristics and machine learning algorithms to detect liveness, incorporating range-checking and binning to enhance spoof detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional facial recognition systems are used, then the system is simple and cost-effective, but the system is vulnerable to spoofing attacks and cannot reliably detect liveness
Solution Approach 1:
The patent transitions from 2D image analysis to 3D geometric analysis by creating point clouds from depth frames. This dimensional change enables the system to capture spatial relationships and surface contours that are invisible in 2D images, providing inherent resistance to spoofing attacks while maintaining computational efficiency through algorithmic processing of the 3D data.
Solution Approach 2:
The system changes the analysis parameters from traditional 2D pixel-based features to 3D geometric features including point cloud density, surface curvature, and spatial distribution. These parameter changes enable reliable liveness detection by capturing the inherent three-dimensional structure of human faces, which cannot be replicated by flat spoofing materials.
2Measurement precision
If depth cameras and advanced processing are used to improve spoof detection, then liveness detection accuracy improves, but computational resources and processing time increase
Solution Approach 1:
The patent extracts only the most discriminative geometric features from the complete point cloud data, such as local surface curvatures and spatial relationships between key facial landmarks. By selecting and processing only these critical features rather than all possible 3D parameters, the system achieves high liveness detection accuracy while maintaining efficient processing speeds suitable for real-time access control.
Solution Approach 2:
The system processes a subset of point cloud data at high precision while using simplified processing for other portions. Specifically, the patent applies detailed geometric analysis to critical regions (such as eye sockets, nasal bridge, and lip contours) while using coarser analysis for less discriminative areas, optimizing the balance between detection accuracy and processing throughput.
Data Source
AI summary
The invention provides, in some aspects, a method of access control that includes receiving one or more frames (“depth frames”) of a depth image stream acquired by a depth camera and, for each of those depth frames, generating one or more scores of the liveness of the image and/or of a candidate individual therein. The method includes creating a point cloud from a respective depth frame and generating the liveness score from a vector based on geometric characteristics of eigenvalues of point-wise neighborhoods within that point cloud. That vector, which characterizes the respective depth frame and surface contours of a face of the candidate individual represented therein, can serve as a measure of how that surface varies and, thereby, permits it to be distinguished from a spoof, e.g., a depth image of faces printed on paper or shown a digital display.