3D face recognition

What is 3D face recognition

3D face recognition is a biometric identification technology that uses a three-dimensional model of the face to establish identity. Unlike 2D recognition, which works with a flat image and is vulnerable to attacks using photographs, 3D recognition analyzes the depth, shape and volume of facial features, which makes it significantly more accurate and resistant to forgeries (photographs, masks, video recordings). The technology is actively used in access control systems, for unlocking smartphones (Apple Face ID, Face Unlock on Android), at checkpoints, in video surveillance systems and in ATMs with biometrics. In Russia, the technology is used in the Unified Biometric System (UBS) and at critical information infrastructure (CII) facilities to ensure secure access to critical objects and information.

How 3D face recognition works

The 3D recognition process includes several sequential stages that ensure high accuracy and protection from forgeries. The first stage is the capture of a 3D model. Structured light, laser projection (for example, Apple's TrueDepth technology using a dot projector and an infrared camera) or stereo cameras are used to build a three-dimensional map of the face. The second stage is the extraction of key points. The system determines up to 80-100 control points and anthropometric landmarks: the distance between the eyes, the shape of the cheekbones, the depth of the eye sockets, the shape of the nose, the depth characteristics. The third stage is the creation of a biometric template. On the basis of key points and the depth map, a unique mathematical model is built — a biometric template that usually occupies from 2 to 5 KB. The fourth stage is comparison and verification. During authentication, the system builds a current 3D model of the face and compares it with the saved template, calculating the degree of similarity. The comparison process takes less than 1 second. 3D recognition works effectively even with changing lighting, the presence of a beard, glasses, makeup or a change in hairstyle. It is resistant to attacks using photographs and masks thanks to depth and relief analysis, as well as the use of liveness detection systems. The technology is often integrated with access control systems (ACS) and cryptographic protection tools to ensure comprehensive security.

Advantages of 3D recognition

The implementation of 3D face recognition technology gives organizations and users the following advantages. High accuracy — a minimal level of false positives (FAR less than 0.001%, FRR less than 0.01%), which significantly exceeds 2D methods. Resistance to forgeries — protection from photographs, masks, videos and 3D-printed models thanks to the analysis of depth, skin texture, micro-movements and thermal radiation (liveness detection). Operation in difficult conditions — insensitivity to lighting, makeup, beard, glasses, hairstyle changes or partial overlap of the face. Contactlessness — convenience and hygiene of use. Speed — identification takes less than 1 second, which ensures high throughput at checkpoints.

Use in various spheres

3D face recognition is widely used in various spheres. In smartphones and mobile devices — for unlocking, authorization in applications, confirming payments. In access control systems — for identifying employees when passing through turnstiles, doors and barriers. At checkpoints — for automatic verification of passengers in airports, at borders and stations. In the banking sphere — for logging in to mobile banks, confirming operations in ATMs and branches. In government systems — in the Unified Biometric System (UBS) for receiving state services without a passport. In video surveillance systems — for searching for suspicious persons in a crowd, monitoring access to events.

Protection from forgeries (liveness detection)

Modern 3D systems use comprehensive liveness detection mechanisms for protection from attacks. Depth analysis makes it possible to distinguish a flat image (photograph) from a three-dimensional face. Skin texture analysis checks the presence of pores, micro-wrinkles and other natural signs of living skin. Micro-movements track natural micro-movements of the face (blinking, micro-contractions of muscles, changes in facial expression). Thermal radiation — infrared cameras record the thermal profile of the face, which differs between a living person and a mask. Glare analysis checks the reflections of light from the eyes and skin. Random requests — the system can ask the user to perform an action (turn the head, smile, blink).

Choosing and implementing a 3D recognition system

The choice of a solution for 3D face recognition depends on the requirements for accuracy, throughput and operating conditions. It is necessary to assess the number of users, the speed of passage, the required level of security, lighting conditions and the placement of equipment. 3D scanners differ in range, type of sensors (structured light, laser projection, stereo cameras, ToF), processing speed and accuracy. For government systems and CII facilities, the use of certified equipment and compliance with FSTEC requirements is required. The design service will help determine the optimal architecture of the identification system.