Fingerprint Liveness Determination Using Layered Classifier Ensemble
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current user verification systems face challenges in distinguishing between genuine and fake fingerprints, particularly in preventing spoofing attacks, where fake fingerprints made of various materials can deceive the system, leading to potential unauthorized access.
Innovation Solution
A method involving clustering training fake fingerprint images into groups based on intensity histograms and training layer classifiers for each group, with an overall classifier generated by combining these classifiers based on difficulty levels, to determine the liveness of an input fingerprint image by sequentially passing features through these classifiers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single classifier is used to determine liveness, then the device complexity is low, but the measurement precision of liveness determination deteriorates
Solution Approach 1:
The patent divides the liveness determination task into multiple specialized classifiers, each trained on specific subsets of training data with different characteristics. This segmentation allows each classifier to focus on particular patterns, improving overall accuracy while maintaining manageable complexity through modular design.
Solution Approach 2:
The patent varies training parameters by creating multiple classifiers trained on differently prepared training sets (e.g., different preprocessing, different data subsets). This parameter variation enables the system to capture diverse fingerprint characteristics, enhancing measurement precision without requiring a single overly complex classifier.
2Measurement precision
If more training data is used to improve liveness determination accuracy, then the measurement precision improves, but the loss of time increases due to longer training duration
Solution Approach 1:
The training data is segmented into multiple subsets, each used to train a specific classifier. This allows parallel training of multiple classifiers on smaller data portions simultaneously, reducing total training time while maintaining high accuracy through the collective expertise of multiple specialized classifiers.
Solution Approach 2:
Each classifier is trained on a partial subset of the complete training data rather than all data. This partial action approach enables faster training of individual classifiers while the ensemble of multiple classifiers collectively achieves high measurement precision through their combined decision-making.
3Measurement precision
If multiple classifiers are combined to improve liveness determination accuracy, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The system segments the classification task into multiple specialized classifiers, each handling specific aspects of liveness determination. This segmentation improves accuracy by capturing diverse patterns while managing complexity through clear modular boundaries and specialized functions for each classifier.
Solution Approach 2:
Multiple specialized classifiers are merged into an ensemble system where their decisions are combined. This merging improves measurement precision by leveraging the strengths of each individual classifier while the structured combination method keeps the overall system complexity manageable through systematic integration.
Data Source
AI summary
Disclosed are a liveness determining method and apparatus and a method and apparatus for training the liveness determining apparatus. The liveness determining method includes extracting, by a processor, a feature from an input fingerprint image, inputting the feature into the current layer classifier, inputting the feature into the subsequent layer classifier, based on a determination that an output of the current layer classifier is live, and determining a liveness of the input fingerprint image to be false, based on a determination that an output of the subsequent layer classifier is fake, wherein the current layer classifier and the subsequent layer classifier are respectively trained based on a plurality of training fake images belonging to different groups.