Liveness Detection Using Low-Resolution Differential Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric liveness detection methods face challenges in accurately distinguishing between live and fake biological inputs, particularly in mobile devices with limited resources, where processing high-resolution images is computationally intensive and resource-heavy.
Innovation Solution
The method involves preprocessing phase images by removing noise and enhancing edge regions, generating differential images, reducing their resolution through pooling operations, creating a minimum map image based on these low-resolution differential images, and performing liveness detection using a neural network model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution images are processed for liveness detection, then detection accuracy is improved, but computational complexity and resource usage increase
Solution Approach 1:
The image processing is divided into multiple stages: initial low-resolution filtering to eliminate obvious 2D spoofing attacks, followed by selective high-resolution processing only for regions of interest. This segmentation allows the system to maintain high detection accuracy for genuine faces while reducing computational burden by avoiding full high-resolution processing for all cases.
Solution Approach 2:
The system performs partial high-resolution processing only when necessary - specifically, high-resolution liveness detection is applied only after low-resolution filtering indicates potential genuineness or when spoofing suspicion arises. This partial action approach maintains accuracy for critical cases while reducing overall computational complexity compared to processing all images at full resolution.
2Measurement precision
If high-resolution images are processed for liveness detection, then detection accuracy is improved, but resource usage increases
Solution Approach 1:
The processing pipeline is segmented into energy-efficient low-resolution screening followed by computationally intensive high-resolution analysis only when needed. This segmentation significantly reduces overall resource usage while preserving detection accuracy for cases that require it.
Solution Approach 2:
The system uses a disposable low-resolution processing approach as a first line of defense - quickly evaluating images at low computational cost to filter out obvious spoofs before committing expensive computational resources to high-resolution analysis, thereby optimizing overall resource utilization.
3Measurement precision
If high-resolution images are processed for liveness detection, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The processing is segmented into fast low-resolution preliminary detection and slower high-resolution detailed analysis. This segmentation enables real-time processing for most cases (using low-resolution) while maintaining the option for high-accuracy verification when needed, thus improving overall processing speed without sacrificing accuracy for critical cases.
Solution Approach 2:
The system performs high-resolution processing partially - only for images that pass the low-resolution screening or exhibit suspicious characteristics. This partial high-resolution action maintains detection accuracy for important cases while dramatically improving average processing speed compared to universal high-resolution processing.
Data Source
AI summary
A processor-implemented method with liveness detection includes: receiving a plurality of phase images of different phases; generating a plurality of preprocessed phase images by performing preprocessing, including edge enhancement processing, on the plurality of phase images of different phases; generating a plurality of differential images based on the preprocessed phase images; generating a plurality of low-resolution differential images having lower resolutions than the differential images, based on the differential images; generating a minimum map image based on the low-resolution differential images; and performing a liveness detection on an object in the phase images based on the minimum map image.