Face Matching
Face matching is a biometric technology that compares the facial features of a person with a reference image to confirm their identity. It is widely used in remote identity verification processes.
Face matching relies on artificial intelligence algorithms capable of analysing facial geometry: the distance between the eyes, nose shape, jawline contours, and other characteristic points. This data is converted into a unique numerical vector (facial template) that can be compared against one extracted from an identity document.
In the context of KYC compliance, face matching occurs after the identity document has been captured. The customer takes a selfie or records a live video, and the algorithm compares this image with the photo on the document. A similarity score is calculated to determine whether the two faces belong to the same person.
The reliability of face matching depends on image quality, lighting conditions, and the algorithm's ability to handle natural variations such as ageing, wearing glasses, or changes in hairstyle. Modern solutions achieve accuracy rates above 99%, while incorporating bias detection mechanisms to ensure fairness across different populations.
Regulations
Real-world examples
- 1.During online bank account opening, the customer takes a selfie that is automatically compared with their passport photo to verify they are the same person.
- 2.An insurance company uses face matching to verify the identity of a policyholder during an online claim, preventing identity fraud attempts.
- 3.A car rental service compares the driver's face with their driving licence photo before handing over the keys, automating the identity check at the counter.