Face Morphing Attacks: The Threat to Digital Identity & How to Stop Them

Picture yourself waiting in line to get your morning coffee from your favourite cafe. As soon as you approach the counter, the cashier instead of greeting you with a smile and taking your order, rushes to his phone, starts recording your video and asks everyone to call the police.

Shocked, you ask everyone present what happened but it’s already too late as the cops are here to arrest you under suspicion that you are the notorious hacker who recently stole US$ 2 million from an online bank account. 

When the case was investigated it was revealed that your pictures from social media were hacked by a fraudster who used face morphing to blend your pictures with his own to make a fake Identity to commit the cybercrime that is now pinned on you. 

This scenario vividly explains the sensitivity involved with Facial Morphing which used to be a harmless graphical technique years ago. But today, regulators like NIST are primarily focused on setting the benchmark standards for Morphing Attack Detection for identity verification solution providers to detect and prevent face morphing attacks.

NIST also explains the risk of facial morphing attacks and how the current facial recognition systems are performing to counter facial identity spoofing through morphing. 

Key Takeaways 

• Facial Morphing is a graphic editing technique which has become a threat to digital identities. 
• Face Morphing attacks are employed by fraudsters to fool facial recognition systems. 
• Face morphing detection requires NIST’s benchmark standards to be followed by facial recognition solution providers. 
• There are 5 best practices that can enable Facial Identity Verification solutions to prevent morphing attacks. 

What is Facial Morphing? 

Facial Morphing or simply morphing is a graphical technique in which two or more images are blended to make a new image. The new image being a mixture of previous images closely resembles them at the same time.

It is a sophisticated process that uses graphical animation and image manipulation to generate new realistic face images that are not associated with any identity database thus making it difficult for IDV solutions to identify and detect advanced morphing attempts. 

How Does Facial Morphing Work? 

Facial Morphing has become a weapon of choice for fraudsters to spoof the KYC (Know Your Customer) identity verification systems. Facial morphing is a sophisticated image manipulation technique that creates a highly realistic face image from two or more blended images.

Face morphing has become far easier since AI tools are now easily available on the internet and mobile phones. Morphed images can fool both humans and facial recognition systems. 

Let’s explore the complex facial morphing system:

Types of Facial Morphing 

Linear Morphing 

This technique involved two intermediate images being blended to create a morphed image. Each image with varying weights is combined according to the image’s features lying in the linear position assigned. 

Feature-based Morphing 

In feature-based morphing, facial features of two source images including nose, eyes, lips, ears, eyebrows, etc are manipulated independently and then blended to create a morph. It enables subtle changes in facial features during morphing. 

Mesh-based Morphing 

This technique creates a mesh grid over the facial region in a video still or another image of two images. It manipulates the vertices of the mesh grid to deform the facial structure and is used in graphic designing and animation for realistic image creation.

Cross-Dissolve Morphing 

It is also called texture-based morphing which blends the texture of two facial images by aligning them. 

3D Morphing 

It manipulates three-dimensionality by high-tech 3D representation of the source images. It includes 3D model interpolation between two source image 3D models. The second method in 3D Morphing is the sculpting technique applied to a 3D mesh representing a face.

Face Morphing Attack Detection 

To detect a morphing attack, Identity Verification Solutions play a crucial part through facial verification and facial recognition procedures. Through advancements like generative AI and Biometrics, Facial morphing attacks have become far superior and difficult to detect. To counter this, IDV vendors, photo-ID issuance authorities and fintech firms are highly interested in how AI and Biometrics can be integrated to create a foolproof facial identity system and the one that can detect morphing attacks well before time. 

To guide the identity verification solution providers in enhancing their efforts against identity fraud, the National Institute of Standards and Technology (NIST) updated its FRVT report on February 21st, 2024 and then on March 13th, 2024 adding algorithmic results from the Norwegian University of Science and Technology (toucan-000 and ntnusub-000) and Hochschule Darmstadt (header-006) respectively. 

This report sets the benchmark for KYC Identity Verification solution providers to improve their solutions’ capabilities of morph attack detection. 

Here’s an overview of the updated analysis of the performance of current Face Morph Detection systems by NIST. It is called the Face Analysis Technology Evaluation (FATE) Morph Test.

Explore Facia’s facial recognition solution suite that enhances your business security with cutting-edge morphing detection technology with high precision and accuracy. 

How to Prevent Facial Morphing Attacks? 

Preventing facial morphing attacks requires a combination of technological solutions aligned with the industry’s best practices. For adaptation with the NIST’s benchmark standards, facial recognition systems can enhance their current morphing detection capabilities. 

Here are five effective ways by which facial verification and recognition systems can prevent advanced facial morphing attacks: 

Biometric Liveness Detection 

Integrate a biometric liveness detection system with the existing IDV systems to ensure that the presented facial images are from an alive person rather than static images. Liveness detection can involve analyzing facial movements, blinking, or other dynamic features to distinguish between a live person and a static image. 

Facial Landmark Analysis 

Facial landmark analysis algorithms can identify and analyze key points on the presented face. It can detect unnatural distortions or inconsistencies in facial landmarks and can help identify sophisticated morphing attempts. 

Texture Analysis and Pixel Matching 

Employ advanced image analysis techniques to assess the texture and pixel-level details of facial images. Trained algorithms can modify themselves and keep updating to identify anomalies or irregularities introduced during the morphing process. 

3D Facial Recognition 

3D facial recognition technology is a futuristic approach in digital identity verification, which analyzes the three-dimensional aspects of facial features. It is highly accurate in detecting morphing attempts by considering depth information and preventing attacks that involve 2D image manipulation. It can also be integrated with Biometric Liveness Detection to further enhance the morphing detection capability of an Identity Verification Solution. 

AI Models 

Train Artificial Intelligence models on datasets of genuine and morphed images to develop robust algorithms for morphing detection. AI and Machine learning can adapt to evolving morphing techniques and improve accuracy over time with evolving threat vectors in facial identity fraud. 

Final Word 

Facial Morphing is now posing a greater threat to the integrity and security of digital identities. With advancements and easy availability of Generative AI tools, this threat vector has sharpened its claws where Facial recognition providers must adapt to NIST’s benchmark standards and improve their solutions accordingly.