Facial Authentication 2D vs 3D: which is more secure?

The changing trend of biometric authentication in the digital era.
Technology trend in facial authentication systems
As enterprise leaders continue to upgrade the identity and access management (IAM) systems, the technology is improving, evolving, and expanding at an explosive rate.
  • Gartner predicts that, by 2022, 40 percent of global midsize and larger organizations will use IAM capabilities delivered as software as a service (SaaS) to fulfil most of their needs — up from 5 percent in 2018.
  • The global facial recognition market size was valued at USD 3.86 billion in 2020 and is expected to expand at a compound annual growth rate (CAGR) of 15.4% from 2021 to 2028.

The technology is improving, evolving, and expanding at an explosive rate. Technologies such as biometrics are extensively used in order to enhance security. These are used across various applications, such as access control, attendance tracking, security and surveillance, and others. Biometrics are universal, unique, and measurable and thus can be used to provide security solutions.

Facial authentication technology is a type of image recognition technology that has acquired broad acceptance over time. The increasing adoption of facial authentication technology across various applications is a driving force for market growth

Biometrics are plausibly to influence online IAM ecosystems, payment systems in a way that they give an all-new dimension to digital economy

3D facial authentication is likely to be the trending biometric authentication system used in 2024. Facial authentication, particularly 3D face modelling, is set to play a very big role in these advancements and will be something to look out in the near future

Biometrics for authentication and identification

Biometrics for authentication and identification can be categorized into three categories: morphological, biological and behavioural. (Figure 1)

The consumer biometrics market is driven by morphological method of authentication. Facial authentication being the most prominent as compared to other biometric methods is considered to be a differentiator because of its natural ability to recognise and discern human faces. Additionally, the process is contactless, discreet and socially accepted.

Recognition vs Authentication

There is always a misapprehension while referring to “recognition” and “authentication” as far as face biometrics is concerned. There always has been tendency to use them interchangeably but both the terms mean different. Recognition refers to the identification of user through image-matching. Authentication operates concurrently, but goes beyond identifying the user to also verify them as a real, live human.

Liveness detection has become a fundamental part of the face authentication process.

Evolution of AI over the past has empowered machines to detect and analyse various living human characteristics and traits, thus making the biometric-based authentication systems robust and accurate.

Further, we emphasize on how 2D and 3D imaging techniques influence and differ from each other while being used for facial biometric authentication systems

Face authentication systems: 2D and 3D approaches

Face authentication is a complex biometric process. Here, we try to comprehend these processes in a simplified perspective.

There are two crucial steps in facial authentication process: registration and authentication. The user must register their biometrics (their face) with the system before their face can be authenticated.

Face Registration

Face registration is a 5-step process: face detection, image quality check, liveness detection, measurement, and representation

The first step in the 5-step process is for the system to distinguish face from the background; then, a quality check is performed on the face’s image. Light condition and position of the head play a crucial role in deciding the efficiency and quality of the registration process. Meanwhile, the user experience should be intuitive and responsive so that it is unaffected by strict criteria for image quality. Hence, maintaining the right balance between user experience and image quality is the key.

As soon as the image is captured and passes the quality check, the system runs checks for liveness of the subject. Liveness detection is the most critical step in the authentication process.

Once the liveness detection is passed, the system allocates nodal points to the various landmarks on the face such as edges, curves, angles, lowered and raised points. With the help of these nodes, the system then generates a face print of the human face which serves as an identifier (Figure 3)

Face authentication

Face authentication can only be performed if the face print is available in the system registry stored through the previously explained process of face registration.

Face authentication is a 6-step process where in steps 1 through 4 are as same as that of face registration process (Figure 2). Step-5 involves matching of the detected face with the registered face print. A score of confidence is awarded based on the computations done by the system comparing the nodal points. In step-6, the system makes use of this score to determine whether the authentication is successful or not

Impact of 2D and 3D face models on liveness detection

As emphasized earlier, liveness detection is critical in maintaining the efficacy and robustness of face-based authentication systems. In the current era of rapidly evolving AI technology, systems can become vulnerable to spoofs such as recorded videos, photographs; 3D face artifacts with the capability to blink, movement of lips and similar facial expressions. Also, spoofs can be created with materials which are very close to the texture of human skin. Above mentioned spoofing tools are not difficult to source and thus the security systems can succumb to such vulnerabilities. Anti-spoofing techniques thus play a very major role in the security systems and are smartly integrated in the systems and serve as prime differentiators. Liveness detection is measured mainly on three indicators: motion, texture and life sign.

Motion Analysis

Motion analysis is used to distinguish between a 2D facial photo and a 3D face. The movement of a planer 2D object is notably different from a 3D object. A video is captured in order for the system to conduct motion analysis with the help of optical flow

Texture Analysis

The texture of an actual 3D face is quite different from that of a face image printed on a 2D paper The system performs analysis for print failure and image blur to differentiate between real and fake.

The system must be trained well in order to detect the real face texture. This when combined with the motion analysis offers better results.

Life sign detection

Life sign detection plays a major role in liveness detection of the real face. It can be performed in two ways

  • Motion challenge: The system asks the person to perform various motions using the head to capture motion data and thus compare it with the captured biometric data.
  • Blink and smile: The system prompts the person to perform specific actions like blinking and smile to determine the liveness of the subject

Combining these two aspects would result in a robust system for face authentication using 3D model.

The bottom-line

We are going through a significant transformation and advancements in the area of biometrics and facial biometrics in particular. As the time progresses, the technology continues to evolve and the process of authentication starts to become significantly less challenging.

However, the business use cases for this technology will remain a crucial ethical question, especially if the technology for facial recognition is misused, disregarding personal privacy

Chiplogic Technologies

Maidenhead Concorde Park
Concorde Road
Building 3, 1st Floor
Maidenhead SL6 4BY
United Kingdom

Sign up to Our Newsletter

© 2020 Chiplogic Technologies.