Security Beyond the Surface: Thermal Imaging in Liveness Detection

Both facial recognition and fingerprints are becoming increasingly popular as preferred modes of biometric security, authentication, and recognition. Yet new methods in the domain are being researched, experimented upon, and they are reaching the public eye every day. One of the more compelling ones among these is using thermal imaging in liveness detection and user identification.

But in a world where such methods as face biometrics already exist, what is the place of thermal scans? Do they supplement face scans or replace them? And is body heat a viable way of checking if an input is real in most conventional applications of the liveness detection tech? Let’s talk about this and more in today’s blog.

What is Thermal Imaging?

Thermal imaging works off of the fundamental principle of biological beings that they constantly emit heat, or in other words, infrared radiation. This makes it useful for biometric systems, as they require quick analysis of biometric variables, which thermal imaging provides a reliable mechanism for.

So, thermal imaging is the process of capturing this infrared radiation or energy emitted by living beings through sophisticated devices known as infrared sensors and thermal cameras, and then turning the captured information into a visual representation that depicts the temperature differences across the captured area. 

It thus creates a visible representation of what is usually invisible radiation, making it easy to understand and comprehend for human beings.

How it Actually Works

Thermal cameras detect infrared radiation, and convert it into an electrical signal, which can then be processed by a system to create the ‘thermal image’. So, when a thermal camera is pointed at a certain area, it picks the infrared radiation emitted from the objects within its view and range, converts the data into electrical signals, and then the signals into an image.

In this image, different colors or shades are used to represent the intensity of warmth that an object or a part of a person is emitting. This usually means yellow, orange, and red for increasing levels of heat, and purple, green, and blue for increasing coldness.

Using Thermal Imaging for Identifying People

When it comes to the world of biometric security, thermal images are somewhat unique in that they can be used for both user identification and liveness checks. Let’s talk first about the ways it can be used for identification.

1) As a Standalone Form of Identification

The first way of using this system is as an independent biometric identification method for security and access. This method, in a similar way to retina scans, looks for the blood vessels that live just below the surface of your face as a method of uniquely identifying you.

These blood vessels are thought to be highly unique to each individual. The vessels, while not able to be seen normally, become clearly visible when viewed through infrared light. So, they can be used effectively as a method for individual identification and recognition through a thermal imaging system.

This thermal pattern analysis can be said to be much more secure when compared to simple face recognition as a form of security, as while it is possible to mimic or recreate a person’s face, it would be highly difficult to find a way to recreate the exact pattern of blood vessels behind the face.

2) Combined with Facial Recognition

The same tech can also be effectively utilized alongside the conventional biometric verification method of facial recognition to make for a highly effective identification system. 

So, in such a process, the user would go through the usual facial scanning process, where a camera scans their face, extracts features, and matches against a database, but at the same time, an infrared camera would also perform vein pattern recognition in the face for added security. This would ensure absolute security, and work as a compensator for the potential weaknesses in a pure facial recognition system.

Recent studies have also shown that conventional facial recognition algorithms might also be effectively utilized for detecting faces in images captured through thermal cameras, without re-training or additional training.

Thermal cameras are more easily able to detect individuals through what might normally obscure a general-purpose CCTV camera, like fog or other similar obstructions. This is why they are commonly used in high-security spaces like defense installations and other military applications. So, thermal scanning combined with facial recognition systems would be a highly effective system for such applications.

A Different Way of Detecting Liveness

For both the two most popular forms of biometric inputs, facial scans, and fingerprint scans, thermal images can also be utilized as an effective form of liveness check. While for fingerprints this takes the form of finger blood flow detection, for facial scans, it directly utilizes body heat.

As stated at the start, thermal images work off of the principle that all living things emit heat and thus infrared radiation. This characteristic of living beings also works as a great method of checking whether an input provided to a facial biometric verification system is from a live individual or a spoof, as an image or spoof would of course not have a proper thermal signature.

Thermal Pattern Analysis

So, during a thermal images-based liveness detection process, a special infrared scanner would capture the levels of thermal radiation emitted by the (claiming to be real) user. It would then convert it into an electric signal which is then displayed as an image.

At this point, the system would analyze whether the data is representative of the kind of thermal output that an actual human face emits. Such an algorithm would of course, much like a conventional liveness detection system, have to be trained on large datasets of images where human thermal patterns are marked and represented.

Thus, based on this analysis,  the system would be able to detect whether an input is from a live person or a spoof. This is how a thermal imaging system can be used to effectively perform presentation attack detection.

When is Thermal Imaging a Good Idea to Use?

As we have discussed, thermal imaging has applicability both as a form of user identification, through vein pattern recognition, as well as a form of spoofing prevention, through face temperature analysis. That should mean that using this technology is a no-brainer, right?

Well, there are a few points that have to be taken into account, and that makes it so that this kind of system is not as easily suited for widespread use as one might think.

Some Major Considerations

A major factor that needs to be considered during the application of this tech is that thermal imaging requires special equipment to be utilized. Infrared or thermal cameras are highly sophisticated pieces of equipment, and that means they are quite expensive. Compare that to conventional facial recognition, which can be done with normal, visible light cameras that already exist in most such places (think phone cameras, CCTV cameras).

A second consideration is that since the algorithms that such a system will utilize work off of even more complex data than conventional detection ones, they require much more intricate development and work to make them both accurate and reliable. While a much more complicated and technical algorithm of course does mean higher cost, it also means not as widespread availability of these algorithms.

So Where Do We Use It?

Keeping these things in mind, the special equipment required, the cost considerations, and so on, thermal imaging for biometric fraud prevention seems best utilized in highly sensitive and high-security use cases, like in high-value financial transactions, or the military as shown above.

These institutions are best equipped to utilize the advanced capabilities that such a system offers. They can afford the high installation and maintenance costs of such a system, and can also take advantage of the advanced accuracy and security that such a system provides, as these spaces often require much more security than normal.

Generally speaking, the special equipment requirements associated with thermal imaging-based biometrics alone make it difficult at best to recommend for widespread public use. Add to that the fact that increasingly large amounts of activity that require user identification and liveness checks are moving online (meaning remotely), it makes it even more unlikely that such tech would be able to be used. 

Thus, while a highly secure form of biometrics, thermal imaging is likely to remain a niche use case, limited to implementation in physical, high-security environments that would be able to better utilize its benefits. To learn more about the best and most widely applicable solutions for biometric security, contact Facia today.