Biometric Enrollment Fraud Detection via Multi-Modal Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric data acquisition systems are vulnerable to fraud, particularly during enrolment, where malicious individuals may use fake fingers or swap fingers to deceive the system, making it difficult to detect fraudulent activities effectively.
Innovation Solution
A method that uses a fingerprint scanner to acquire initial biometric data and a camera to capture context images of the hand and its environment, comparing the data to ensure the fingerprints belong to the same individual and detecting fake fingers by analyzing the hand's outlines and environment, thereby validating the integrity of the biometric data acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only fingerprint scanner is used to acquire biometric data, then the acquisition process is simple and fast, but the system is vulnerable to fraud using fake fingers or swapped fingers
Solution Approach 1:
The patent combines a fingerprint scanner with a camera system to create an integrated acquisition device. The fingerprint scanner captures ridge patterns while the camera captures the hand's overall shape, position, and context. By merging these two sensing modalities into one system, the patent enhances reliability through multi-modal verification while keeping the device complexity manageable through integrated design.
Solution Approach 2:
The patent transitions from two-dimensional fingerprint scanning to three-dimensional verification by adding contextual imaging. The camera captures the hand's spatial configuration, finger positions, and environmental context, adding a dimensional layer of verification that detects fake fingers and swapped fingers through their physical impossibilities in 3D space rather than just 2D pattern matching.
2Reliability
If operator monitoring is used to detect fraud, then fraud detection capability is improved, but the system remains vulnerable to accomplice operators and is not fully automated
Solution Approach 1:
The system performs self-verification by automatically comparing the captured hand context against expected anatomical configurations and finger positions. The automated algorithm detects inconsistencies such as fake fingers (which lack proper blood flow, temperature, or structural continuity) and swapped fingers (which create impossible hand geometries), enabling the system to detect fraud without human intervention and achieve full automation.
Solution Approach 2:
The system incorporates feedback loops where the camera images are continuously analyzed and compared against the fingerprint data and expected hand models. When inconsistencies are detected (such as a finger that doesn't match the hand's overall configuration or shows signs of being artificial), the system automatically rejects the enrollment attempt, providing real-time feedback that eliminates the need for operator judgment.
3Productivity
If only a small number of fingers are acquired for each individual, then the enrolment process is faster, but it becomes very complicated to identify fraudulent individuals during analysis
Solution Approach 1:
The patent performs preliminary verification during the enrollment process itself by capturing and analyzing the hand's contextual information before storing the fingerprint data. The system validates the hand configuration, finger positions, and overall anatomy upfront, ensuring that only legitimate enrollments proceed to data storage. This preliminary action prevents fraudulent data from entering the database, making subsequent identification straightforward regardless of how many fingers were enrolled.
Data Source
Figure 1
Figure 2
AI summary
A method of verifying acquisition of biometric data, the method comprising: - a first acquisition step (E1) comprising acquiring first biometric data; - a second acquisition step (E2) comprising using a camera (6) to acquire at least one context image comprising a first image portion representative of the same morphological characteristic and a second image portion representative of an environment of the morphological characteristic; - an extraction step comprising extracting second biometric data representative of the morphological characteristic from the first image portion; and - a verification step comprising comparing the first biometric data with the second biometric data and analyzing the second image portion.