Face Morphing Online and its Threat to Digital Security
Face morphing on digital platforms is a new and unique method of identity masking that started as a fun novelty, but over time has become one of the more sophisticated ways to commit many different kinds of frauds, theft, and scams online.
The increasing ease provided by websites and platforms that allow face morphing online has added a layer of concern and problems to this increasing criminal activity and practice.
In this article, we shall discuss in detail how this threat manifests in today’s digitally connected world.
What is Face Morphing
Face morphing is an image/video manipulation technique that can be used to blend or mold a different face to create a new face or image. It is a sophisticated mechanism that uses two input faces to create a new third output face that is close enough in terms of features and appearance to the input faces that it can trick a decent amount of security software.
How does it work?
A facial morphing attack is done using a very long and complicated process involving many steps.
The first step is to take the two input images and establish and mark the corresponding features that make up each face like the eyes, ears, mouth, and other fundamental landmarks. This helps to identify how the basic characteristics of the two images compare with each other in terms of space. Advanced morphing tools even take into account things like wrinkles and facial hair.
The second step creates a mesh on the corresponding points of the two images. This helps the software to analyze how the two images can best be morphed together to form a cohesive new image.
The third step involves warping the image according to the previously created meshes. In this process, usually, the first image is morphed and warped so that its features match the shape of the corresponding features on the second image.
The fourth step involves a blending process. Here, the visual details of the two images, including colors and skin texture, are cross-dissolved or blended in a way that a cohesive image is created.
The final step involved is rendering the created face into the required image or video sequence, the final image or face is now created.
An additional step might happen where a graphic design professional might do manual touching up and refining of the face to make it look more realistic. This can help to remove any artifacts or any other sort of digital trace that might identify the face as being artificial intelligence generated, reducing the chances of morphing attack detection.
The Threat Posed by Face Morphing
The ‘double-edged sword’ philosophy applies to this form of tech just as much as it does to any other new digital advancement coming out at the moment. A face-morphing attack can have serious consequences and create real problems that often have to be dealt with on an urgent basis.
The main application basis for the malicious use of facial morphing techniques is in the forgery of identification documents and materials. The creation of a unique and realistic face in the creation of fake identities can make for a more intricate forgery. These attacks thus tend to be targeted towards facial verification-based identification systems and procedures, both manual and digital.
The point of concern practically speaking of a morphed image being used in an identification document or passport is that two different individuals (the two whose faces were used to make the image) can share and use a single document. This can lead to much illegal activity including illegal migration and border crossing in particular, but also fraudulent financial activity and more.
How to Detect and Prevent Morphing Attacks
With the serious concern that has formed regarding the potential damage of a face deformation attack occurring, different techniques and technologies have been made to try to combat this issue.
Liveness Detection
One of the most effective ways to detect a face morphing attack is facial liveness detection software. This kind of tech works to detect whether the input provided to the system is from a live individual, or if it is digital/computer generated.
Biometric liveness detection can work to analyze facial aspects that indicate natural human movements like blinking and other micro-movements to identify between a static image and a live person.
Specially trained artificial intelligence identification software able to detect abnormalities in facial features can be useful in identifying any form of manipulation of facial characteristics. This kind of system might be trained against a large dataset of real and morphed faces, making it highly accurate in detecting abnormalities.
Conclusion
What started as simple entertainment software has been taken to be used by individuals with malicious intent to carry out illegal activities. Easier access has only elevated the threat posed by this form of software, requiring extensive development and advancement into identity verification tools to combat this attack.