Soft Biometrics: Gait Recognition and Keystroke Dynamics

Humans are formed up of patterns, and everything they do is based on them. Machines are taught behavioural patterns that people use to do any activity in order to teach them human behaviour. You must be familiar with biometric authentication, such as fingerprint or facial recognition, which smartphones conduct for security purposes. 

 

This is accomplished through the use of behavioural biometrics including gait patterns and keystroke dynamics. In this post, you will learn about pattern recognition techniques that aid in the recognition of human identity.

 

But before delving deep into the techniques, it is very important for you to understand the significance of this type of pattern recognition.

 

The significance of soft biometrics:

 

There might be a lot of questions regarding how recognizing and implementing these soft biometric patterns can help us. Here we have stated five applications of pattern analysis and recognition:

 

1. These give a more tailored and improved customer experience:

 

When joining up for an information technology-enabled service, consumers frequently avoid sharing too much information. Furthermore, persons with impairments may find it difficult to begin utilising a digital platform without entering too much information. In such circumstances, a computerised estimate of soft biometrics might be beneficial.

 

For a smooth and personalised experience, businesses may adjust their platforms and services to the demographics of their users. Additionally, predicted soft biometrics can be used to restrict access to specific resources or platforms.

 

2. These improve the brand's recognition rate:

 

By introducing inferred soft biometrics such as age, gender, weight, and height into the authentication and identification pipeline, the performance of authentication and identification systems may be enhanced.

 

3. These aid in the detection of fake social media profiles:

 

Phoney accounts and fake news are spreading throughout social media platforms. Individuals that pretend to be someone else, such as a different gender, height, age, or career, are not rare. Individuals' validity might be difficult to evaluate depending on the sort of information they publish. Soft identifiers based on the typing pattern that is precisely approximated can assist in detecting these profiles and taking necessary action.

 

4. These help in targeted advertising:

 

Soft biometrics may be used by businesses to target persons of a given height, weight, gender, and age group who are more likely to be interested in a product than the general public.

 

5. These help in Forensics:

 

Individual identification has never been more important than it is now, as the volume and kind of cybercrime continue to grow. Soft biometrics can aid in the early detection of these crimes, saving victims from losing a lifetime's worth of earnings.

 

Also Read | 5 Applications of Bioprinting in Pharmaceutical Industry

 

What is Gait Analysis?

 

The systematic study of human movement is known as gait analysis. The examination and evaluation of the gait patterns and features of the person/suspect, and comparing these aspects with the scene of crime evidence for criminal/personal identification, is characterised as forensic gait analysis or forensic gait comparison. The sequence of footprints collected at the crime scene and the closed-circuit television camera (CCTV) film are the general sources of evidence for forensic gait analysis. Footprints are one type of evidence that may be found at a crime scene.

 

There are three types of Gait recognition:

 

1. Vision-Based Gait recognition:

 

The goal of vision-based gait recognition is to recognize a person using multiple characteristics collected from a video sequence of them walking. Vision-based gait identification has the benefit of being inconspicuous when compared to other biometric elements such as fingerprints. Because an individual must demonstrate specified behaviour (e.g., staring in a specific direction or remaining motionless for a recognition delay period) to be identified, iris recognition systems have an intrusive interface. Vision-based gait recognition systems, on the other hand, do not require any individual touch other than walking.

 

The majority of existing vision-based techniques analyse silhouettes in sequences of photos of walking human people. They may be categorised into two categories: Model-based analysis and Model-free analysis. 

 

The model-based analysis seeks to explicitly model the human body or motion, and it generally involves doing model matching in each frame of a walking sequence in order to quantify parameters like trajectories on the model. Based on the prediction or estimation of characteristics relating to location, velocity, form, texture, and colour, model-free analysis builds a correlation between subsequent frames.

 

2. Floor-based Gait Recognition:

 

The biometric sensor utilised in the vision-based technique, a camera, is the major distinction between the floor- and vision-based systems. Vision-based gait recognition has a significant advantage over other contact-type biometric recognition systems like iris recognition because of its unobtrusive and transparent interface. 

 

Unfortunately, because it is subject to environmental elements such as shadow and light intensity, a camera is not always the best option. Furthermore, video surveillance might jeopardise a person's privacy. 

 

In vision research, such an intrusion in privacy-sensitive areas like one's home has been criticised. A floor, on the other hand, is employed as a biometric sensor in the floor-based method, which collects the different aspects of one's stride. As a result, the floor-based method not only ensures individual privacy but also resists shadow and light impacts.

 

3. Portable Sensor-based Gait Recognition:

 

The portable sensor-based technique, in which accelerometers are commonly utilised as biometric sensors, is the most modern approach in gait identification. The accelerometer-based technique, like the floor-based approach, offers various advantages. 

 

One advantage is that it offers a non-obtrusive user interface that does not violate a user's privacy. Users must, on the other hand, carry or connect an accelerometer and a motion-recording device to their bodies in order to be identified. 

 

Accelerometers connected to various regions of the body, such as the hip, lower leg, or waist, assess acceleration signal characteristics while the user is walking in the sensor-based technique. The output of device accelerations in vertical, forward-backward, and sideways directions is used to extract features.

 

How does Gait Recognition Work?

 

Gait recognition technology collects data from a variety of sources or capture devices, such as video cameras, motion sensors, and so on. The data is processed through a series of recognition stages. The core algorithm identifies contours, silhouettes, and specific human traits, as well as processes the supplied data. The feature extraction algorithm then takes over, allowing one stride to be distinguished from another. These algorithms might be diverse, as well as their needs. Some algorithms, for example, are intended to interpret visual information, while others employ sensor data.

 

The four components of the most typical gait recognition system are as follows:

 

• Data collection on gait:

 

A gait can be captured using video cameras or wearable sensors. The unique clothes that performers wear on set so that motion artists may subsequently design the character based on their motions may be the most noticeable example of such sensors. 

 

Another approach of gait capture is to utilise radar to detect moving objects from a distance. Radio waves are used to irradiate the item of interest, which then reflects off their body. The technology detects reflected waves and utilises the information to identify the source.

 

• Silhouette segmentation:

 

For investigations that employ video camera recordings, this stage is acceptable. Vision-based algorithms examine a binary picture of a person's silhouette retrieved from the footage. Silhouette segmentation aids the algorithm's processing and mapping of a whole image.

 

• Detection of contours and crevices:

 

The technology then specifies the human body's limits, emphasising the contours. Depending on what technology (cameras or wearable sensors) is utilised to record the gait, the strategies employed to achieve this aim may differ.

 

• Feature extraction and classification:

 

The individual aspects of a gait are decided in the final phase. The person is identified using a classifier, which is subsequently placed into a database and utilised for detection.

 

Also Read | 3D Bioprinting - Applications, Advantages and Disadvantages

 

What is Keystroke Dynamics?

 

The study of keystroke dynamics biometrics has grown in popularity in recent years. The fundamental reason for this initiative is that keystroke dynamics biometrics are cost-effective and can be simply incorporated into current computer security systems with few changes and user participation.

 

Keystroke recognition is a behavioural biometric that uses a person's distinctive typing style to validate their identification. Typing patterns are generally collected from computer keyboards, although the data might be acquired from any input device with conventional keys that have a tactile response (e.g., cellular phones, PDAs, etc.). 

 

Although other measures are possible, keystroke dynamics patterns are mostly obtained from the two events that comprise a keystroke: the Key-Down and Key-Up. The Key-Down event happens when a key is first depressed, and the Key-Up event occurs when the key is then released. The intra-key and inter-key time variances between these occurrences are then used to determine a variety of unique properties.

 

How do keystroke dynamics work?

 

The biometric template used to identify an individual using keystroke dynamics is based on the typing pattern, rhythm, and speed of typing on a keyboard. Dwell time and flight time are the basic measures for keystroke dynamics. Dwell time refers to how long a key is held down, whereas flight time refers to how long it takes to release a key and press the next one. 

 

When typing a string of characters, the time it takes the subject to find the proper key (flight time) and the time he keeps a key down (dwell time) are both unique to that subject and may be calculated independently of the overall typing speed. The speed with which specific character sequences are written might vary greatly from person to person. Someone who is used to typing in English, for example, will be faster at typing specific character sequences like 'the' than someone with French roots.

 

Here are a few components of keystroke dynamics:

 

• Data acquisition:

 

This is the first stage, in which raw keystroke data is acquired using various input devices. A standard computer keyboard, a customised pressure-sensitive keyboard, a virtual keyboard, a particular purpose num-pad, a cellular phone, and a smartphone are examples of these.

 

• Feature extraction:

 

After that, the raw keystroke data is processed and saved as a reference template for future use. Before feature extraction, certain preprocessing processes may be used to assure or improve the quality of feature data.

 

• Classification/Matching:

 

Most recognition systems are built around this phase, in which feature data is classified and discriminated for subsequent use in decision-making. Previous studies have used a wide variety of methods with the common objective of improving authentication accuracy.

 

• Decision and retraining:

 

The system receives the claimant's feature data and compares it to the reference template using classification techniques. To evaluate if a user is authentic or not, a final judgement will be made based on the results of a classification or matching algorithm.

 

Due to the variety of the user's typing style, it is required to retrain the stored reference template on a regular basis to reflect the current changes.

 

Advantages of Keystroke dynamics:

 

Here are some advantages of keystroke dynamics:

 

1. The software can monitor keystroke events to millisecond precision. As a result, replicating one's typing sequence at such a high resolution without exorbitant effort is impossible.

 

2. In contrast to typical physiological biometric systems that rely on specific device and hardware infrastructure, such as palm print, iris, and fingerprint recognition, keystroke dynamics recognition is fully software implementable. The advantage of reduced reliance on specialist hardware not only lowers deployment costs but also presents a perfect condition for use in a remote authentication environment.

 

3. The degree of transparency provided by keystroke dynamics biometrics is a big advantage it offers over other alternatives. Because the capturing of keyboard patterns is done via backend software implementation, it needs no or little changes in user behaviour. This simplicity benefits not just system designers, but also end-users with little or no technical knowledge.

 

4. Researchers have discovered keystroke dynamics biometrics as a possible option that can offer an extra layer of security while also extending the life of a password. The possibility to combine the simplicity of a password system with the greater reliability of biometrics is provided by keystroke dynamics biometrics. The use of keystroke dynamics biometrics allows users to concentrate on establishing a secure password without being overwhelmed by several password sets.

 

5. Continuous monitoring and authentication are sometimes overlooked, despite their importance. Biometrics based on keystroke dynamics provide a mechanism to continually verify a user's genuine identification. Keystroke patterns can be continuously examined and reevaluated as long as user engagement with the system via input devices continues.

 

Also Read | 5 Applications of Nanotechnology in Biology

 

Conclusion:

 

With the growing volume of data, it's become even more important to improve security. Soft biometrics are a terrific approach to making any gadget more secure. We've focused on two methodologies in this article: gait analysis and keystroke dynamics.