Identity Verification Anomaly Detection via Background Image Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing identity verification technologies struggle to detect high-quality image forgery and the use of 'front persons' in online services, leading to fraudulent account registrations that bypass KYC and AML restrictions, which are often part of digital banking and fintech services.
Innovation Solution
A method using a model to generate descriptors of visual features not associated with a user's face or identification document, comparing these descriptors across verification requests to identify anomalies, such as similar image sources or locations, thereby detecting fraudulent serial attempts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing identity verification technologies analyze facial features or identity documents to detect forgery, then they can identify obvious fakes, but they fail to detect high-quality image forgery and front persons
Solution Approach 1:
The patent extracts and analyzes background elements from identity verification images rather than focusing on the person's face or document. By taking out the background information (location, lighting conditions, objects around the person), the system can detect patterns that indicate fraud without being fooled by high-quality forgeries or front persons who present themselves legitimately.
Solution Approach 2:
The background serves as an intermediary element that indirectly reveals fraud patterns. Instead of directly analyzing the person's identity (which can be manipulated), the system uses background features as a mediator to infer whether the verification is legitimate, since fraudsters often reuse the same backgrounds across multiple verification attempts.
2Ease of operation
If the system analyzes only the person's face and document in identity verification, then verification is simple, but fraud detection capability is insufficient
Solution Approach 1:
The patent segments the identity verification image into distinct components: the person's face/document and the background. By analyzing these segments separately, the system maintains the simplicity of direct verification while adding fraud detection capability through background pattern recognition. The background analysis is performed as a separate segment that complements the traditional verification process.
3Ease of manufacture
If high-quality image forgeries are used, then identity verification is successfully spoofed, but the system lacks the capability to detect such sophisticated fakes
Solution Approach 1:
Instead of trying to detect the person's identity accuracy (which is vulnerable to high-quality forgeries), the system inverts the approach by analyzing the background environment. The background remains consistent across legitimate verifications but varies when different locations are used, providing a reliable detection mechanism that is not affected by the quality of facial or document forgeries.
4Productivity
If the system processes each identity verification request independently, then processing is efficient, but serial fraud attempts using the same templates are not detected
Solution Approach 1:
The patent merges the analysis of multiple identity verification requests by extracting and comparing background features across different requests. Instead of processing each request completely independently, the system combines background information from multiple requests to identify patterns and serial fraud attempts, maintaining processing efficiency while enhancing fraud detection reliability through cross-request analysis.
Data Source
AI summary
The technical solution aims to verify digital identity and more particularly to verify digital identity by online proofing.A method for detecting anomalies in identity verification performed by a processor comprises the following steps (FIG. 1, FIG. 2): receiving an identity verification request comprising image data (10), wherein the image data contains a person's face; generating an image data descriptor (11) using a model (20) configured to determine a set of visual features not associated with the person's face in the input image data (10); searching for image data descriptors similar to said generated image data descriptor (11) among image data descriptors belonging to other identity verification requests; in response to finding at least one similar image data descriptor, marking said identity verification request as anomalous, otherwise marking the request as not anomalous.