How Many Types of Biometrics are there?

What are some of the spontaneous examples a layman may give when asked about “biometrics”? Most people will start with fingerprint recognition, face recognition, iris recognition, etc. and a few of them may starch it to DNA recognition, retina scan, hand geometry and may be a couple of more, if they are familiar with the subject matter.

Despite the common perception of biometrics being limited to only a few popular recognition methods, the number of biometric methods are way more than people generally know. This article lists some common and not-so-common biometric methods. Some of them may even surprise you with their existence.

DNA matching

We have seen it in crime thrillers and CSI stories. Some part of a suspect’s body like hair or blood picked by forensic folks from a crime scene is taken through all dramatic lab investigations and finally help catch the criminal. No matter how dramatic it may look on movies, DNA profiling or DNA fingerprinting a proven method of personal recognition, in which a physical sample of an individual can be analysed to confirm his/her identity. DNA profiling has always been an integral part of forensic investigations. When done for non-forensic purposes, a DNA sample is collected with the subject’s consent. DNA sample is usually collected through a buccal smear, however, blood, semen, saliva, etc. may also be collected and used for DNA profiling.

The collected sample is taken through one of many methods available for DNA profiling like RFPL Analysis, PCR Analysis, STR Analysis, AmpFLP, etc. The results are then compared against other samples to find a match. DNA profiling has been considered to be a very accurate method of identification; however a recent study undertaken by Israeli researchers showed that DNA can be manufactured in a lab and can be used to manipulate the results.

Image: A U.S. customs and border protection analyst is examining a DNA profile to determine the origin of a commodity. (Photo Credit: Wikimedia Commons)

Pros

• Very accurate method of identification.
• Having been used in law enforcement for quite some time, it has become a matured and reliable method of identification.
• Unlike many other biometric identification methods, it does not suffer from system performance issues.

Cons

• Unlike most other biometric recognition methods, which can collect biometric data with imaging techniques, DNA profiling requires physical sample.
• Process of DNA fingerprinting is complex and most part has to be executed manually, unlike most other biometric identification methods which are processed automated systems.
• Being biological samples, they need to be stored with appropriate environmental conditions, unlike most other biometric samples, which can be stored on digital media.
• Samples can be prone to contamination.

Ear acoustic authentication

External shape and size of human ears help them acquire sound waves and route them through the ear canal. Shape of the external ear (also called pinnae or auricular) is considered a physiological biometric characteristics, however, ear acoustic authentication takes ear based biometric recognition to an entirely new level. Developed by NEC with the cooperation of Nagaoka University of Technology, ear acoustic authentication uses sound waves to determine the shape of ear canal of an individual. Shape of human ear canal is considered to be unique for each individual like fingerprints or iris patterns, though there is no evidence behind this consideration (which is also the case with most other biometric methods).

To be able to authenticate with this ear acoustic, the subject has to wear special earphones that also include a microphone to capture sound waves reflected from the ear canal. A high frequency inaudible sound wave (18 to 48 kHz) is used for the purpose. Since each individual features a different shape of ear canal, reflected wave differs for each individual. Microphone captures this reflected wave and the recognition system can identify the person on the basis of pre-established identification information.

Pros

• Unlike most other popular biometrics, which are exposed and can be vulnerable to spoofing, shape of ear canal stays secure inside the ears.
• Authentication can be done on the go, there is no need to put attention to any scan or imaging, making it more suitable for today’s fast paced life.

Cons

• Requires the subjects to wear special earphones, which are external devices.

Eye vein recognition

Human eyes offer a lot of possibility of personal identification with its different structural elements. Be it pattern formed by iris muscles or the capillaries of retina. While iris recognition is possible from a distance, retina recognition can be intrusive, hence uncomfortable for some users. Other than these two, eye vein recognition is another biometric recognition method, which makes use of pattern formed by veins on sclera.

The concept of eye vein recognition was developed by Dr. Reza Derakhshani, an inventor and professor of computer science and electrical engineering at the University of Missouri, Kansas City. He found that patterns of blood vessels in the sclera can be useful for personal identification as it is a unique individual characteristic, given the enormous possibility of variations in individuals.

The technology uses vein pattern of sclera, which is the white part of eyes. These veins are also visible with naked eyes and eye vein recognition technology can capture them with the help of a digital camera. The recognition system processes these digital images to generate a biometric template, which can be associated with the identity data of a person. The person can later authenticate his/her identity using eyes. Unlike retina or iris recognition, the subject has to glance either side to get verified.

EyeVerify, which is a biometric solution provider, offers this technology with its products for applications like mobile authentication for many banking and financial service providers.

Pros

• Eye vein patterns do not change with age, eye redness or alcohol consumption.
• The technology works even with glasses or contact lenses.

Cons

• Being a patented technology, it is offered as a commercial solution for mobile authentication.
• Mobile phones should be held close to the eyes to enable the system to scan vein pattern.
• May not be available for older smartphones or phones without camera.

Facial recognition

Facial structure is the primary approach that humans leverage to recognize others. It is also a well exposed biometric modality that can be used for human identification and authentication. Now when we have high quality cameras with ability to zoom in multifold, it is also possible to identify a subject from a distance, making facial recognition also suitable for security and surveillance applications.

Facial recognition is easy to setup. A digital camera and facial recognition software is all you need to setup basic facial recognition ability. For security inclined applications, more hardware like infrared light emitter and camera, multi-camera setup, etc. may be deployed.

Facial recognition is one of the rapidly growing biometric modality. Rise of smartphones and personal computing devices is one of the reasons behind this growth. Modern day phones mandatorily carry more than one camera (front and back), which makes it easy to leverage face recognition for user authentication. Now device makers are also including additional hardware for face recognition like Infrared emitter and a camera to capture IR illuminated 3D map of facial structure. This setup was first introduced by Apple with iPhone X with FaceID moniker.

Pros

• Easy to setup and no additional hardware needed for most of present day computing devices and smartphones.
• Can be used for personal recognition as well as surveillance applications, making face recognition a versatile biometric modality.

Cons

• Basic face recognition systems may be vulnerable to manipulation and imposter attacks.

Finger vein recognition

Finger vein recognition is a biometric recognition method which makes use of unique pattern of blood vessels found beneath the skin of the human finger. This pattern is formed by veins, the blood vessels that carry blood towards the heart. Since there is not definite way how an individual’s blood vessel will form the pattern in the finger (or palm, or pretty much everywhere), it has very high possibility to differ significantly among individuals. This unique pattern formed by veins in human finger can be used for personal identification and authentication.

VeinID, an authentication technology based on human finger vein pattern was developed and patented by Hitachi in 2005. VeinID authentication system can match the vascular pattern of human finger with other already or newly processed finger vein IDs. VeinID recognition system makes use of near infrared light and a monochrome charge-coupled device (CCD) camera to obtain the vein pattern.

As near infrared light passes through the finger, it is absorbed by deoxygenated blood haemoglobin present in the veins and the image is captured by a CCD camera. The finger vein pattern appears as dark lines in the resulting image. The recognition system processes and digitizes this data which is sent for user enrollment or verification as the case may be.

Pros

• Hidden beneath the finger skin, vein pattern is unexposed and can only be scanned with special setup, hence hard to replicate and circumvent the system.
• Being located beneath the skin, finger vein pattern is securer than fingerprints, face recognition and similar recognition methods in which biometric characteristics are exposed and can be collected without a subject’s consent.

Cons

• The technology is still improving and only a very limited numbers of applications / equipments are available.

Fingerprint recognition

Fingerprint recognition is considered to be the oldest and most developed biometric recognition method. Since the early days of fingerprint identification, it has been a part of forensic investigation. It kept on improving as the process of fingerprint identification and matching advanced in forensic investigation agencies with the use of technology. Today, fingerprint recognition is used from mobile devices to door locks and even for high security access control. Identification and authentication on the go have been made possible by tiny yet efficient fingerprint sensors for mobile devices.

Fingerprint recognition systems rely on the unique ridge pattern found on the fingertips of people. This pattern can be collected using different techniques like optical, capacitive, thermal, etc. The captured image of human fingerprint is enhanced to make it usable and then taken through sophisticated algorithms to produce a digital biometric template, which is unique to the subject. When identity data of an individual is associated with this template, it becomes biometric fingerprint identity. This biometric template can be compared against existing or fresh scans and the biometric system returns a match or no match, as the case may be.

All the process described above is what goes underneath a fingerprint recognition system, at user level, it is just a “touch and go” process, in which getting access is as easy as touching the sensor.

Pros

• Inexpensive (yet secure), easy to setup and most developed biometric modality.
• Wide deployments across all industry types have made it more reliable method of biometric identification.

Cons

• Having been associated with forensic investigations and law enforcement, early implementation suffered with low acceptability.
• Performance of recognition systems suffer with surface condition of fingertips like wet or dirty fingers, scars, skin diseases, etc.

Footprint and foot dynamics

Human footprint is considered to be a unique physiological trait and bears distinctive properties like the palm print and fingerprint. Just like skin ridges on palm and fingers do not change for life, ridge structure of human foot also stays the same throughout a person’s life. This gives us an opportunity to use footprints as a mean of personal identification, just like fingerprints or other biometric identifiers. Underlying technology for scanning and processing footprints stays more or less same as other finger ridge recognition technology, however, footprint scanning technology is still under development and various approaches are being experimented with.

Current systems use a camera to capture footprints and after pre-processing the recognition system extracts shape using gradient vector flow (GVF), and minutiae. When the recognition technology comes close to its maturity, it is expected to use features like foot shape, texture, friction ridge, etc. In dynamic foot print approach, the footprint of a subject on the move is used to identify him/her.

Despite being extensively studies for forensic investigations and being a unique physiological characteristic, footprint biometrics is not commonly used for human identification or authentication.

Pros

• Use of dynamic footprint recognition is considered complicated in commercial biometric systems due to several like user unfriendliness of data acquisition process, etc.
• Can be useful some special use cases like spa, thermal baths and covert identification.
• Since not intended for high security applications, storage of footprint and foot dynamics biometric data does not pose any security threats.

Cons

• Not suitable for high security applications like electronic banking, access control, etc.

Gait recognition

Human gait is a distinguishable feature. We all have some ability to identify a familiar person just by observing his/her gait. With the rise of computer based imaging and machine vision, it has become possible to pass this ability to the computers. Like many other biometric systems, gait recognition systems are pattern recognition systems, which can map pattern of human gait. Each individual has an idiosyncratic and distinctive way of walking which can be mapped using modern computing and imaging. Mapped pattern is digitalized and taken through recognition algorithms designed to generate a digital biometric template unique to the subject involved.

Since gait is a behavioral characteristic and repeats itself in cycles as long as someone moves. The data generated out of this pattern, which is considered unique, can be captured and processed to recognize that individual.Gait recognition is still in its infancy and experts are mainly putting their focus on two approaches:Analysis of video sample of a subject’s gait, which is the most popular and easy to implement method. In this approach, gait pattern is mapped in a mathematical model using gait recognition algorithms.Radar based systems, which records the reflected radar waves out of person’s gait and maps them in the mathematical model to create a digital profile of the person’s gait. This profile, which is also called the biometric template of the person, can be used to identify the person.

Pros

• Unobtrusive method of recognition, making it suitable for biometric law enforcement security and surveillance applications.
• Can identify people from a distance.
• Easy to setup, video footage from existing surveillance cameras can be used, making it inexpensive to deploy.

Cons

• Still in its infancy, hence suffers from low reliability of results.
• Gait recognition’s non-invasiveness raises privacy related concerns.

Hand geometry

Once a dominant method of personal identification, hand geometry has survived ups and downs of biometrics industry. However, in recent years, due to the rise of other more efficient yet cheaper biometric identification methods like fingerprint and facial recognition, its relevance in modern application has decreased.

Hand geometry recognition is one of the longest implemented methods of personal identification. It leverages the idea that each individual has a unique shape of hand, which can be used to uniquely identify him or her. Automated systems, which can capture geometry of human hand; establish identity of the basis of captured data and can verify it later, are called hand geometry readers. At first look, hand geometry readers may look like palm print scanner, however, they server entirely different purpose. Hand geometry readers can capture image of a subject’s hand to measure length of fingers, their thickness, width, curvature and relative location of other features, which are considered to be unique to a person.

The recognition system capture top as well side image of user’s hand. This raw image is used for generating a silhouette, which is further used for the measurement of hand. The system performs at least 90 measurements and analyses 31,000 points of silhouette. The system finally generates a biometric template unique to the subject.

Pros

• Easy to use, simple and fast method of identification. Hand geometry scanners are designed specifically for employee time and attendance applications.
• More reliable than traditional card or paper based systems.
• High level of acceptance.
• Performance of system is not affected by surface condition of skin.
• Hard to circumvent.
• Can withstand harsh environmental conditions like extreme cold or heat, where many other biometric systems will fail.

Cons

• Not suitable for high security applications.
• Diseases, weight loss/gain, injury, age, etc. can change the shape of hand significantly.
• Hand geometry readers can be way more expensive than many other more advanced biometric recognition methods like fingerprints and face recognition.

Iris recognition

Iris recognition is one of the first generation biometric methods which have matured over time and taken over identification and authentication applications for many use cases. In the human eye, iris is the colored portion in the shape of a ring. If you look closely, you will find it is made of many asymmetric thick thread-like structures. These thread like structures are the muscles that help adjust shape of the pupil and only allow appropriate amount of light in the eye.

The muscular folds of iris create a pattern which is considered to be unique for an individual and can be used to identify him/her or verify the identity. Iris based recognition is made possible by iris scanner, a setup of digital or infrared or both cameras along with a software to process the captured iris image. The software may also feature liveness detection with options like a subject is required to blink his/her eyes to have them scanned.

The image is pre-processed by the software to enhance its usability and then processed by recognition algorithm to extract unique features. A biometric template is eventually generated which can be used to enroll a user or perform a match against already enrolled data.

Pros

• Less intrusive than retina recognition, which is another eye based biometric recognition method.
• Modern iris recognition system can identify a user from a distance.
• High level of performance and stable biometric recognition method.

Cons

• Collectiveness of the iris can be affected by age and eye diseases that deteriorate transparency of cornea.

Keystroke dynamics

Passwords alone have been the sole guardian of IT system and information security for a very long time. However password based security is increasingly losing its relevance amid present day threats to the information security. Passwords alone cannot withstand current data security challenges and additional security measures have to be implemented. Keystroke dynamics (also called keystroke biometrics), a behavioral biometric approach, can be the solution.

Keystroke dynamics leverages the fact that people follow a definite pattern while typing on a keyboard or keypad. This keystroke rhythm of a user can be used to establish a biometric profile, which can be used to identify or authenticate him/her. Time taken to press each key, pause between key presses, letters typed per second/minute and several other measures are taken to generate a keystroke profile of a user. Keystroke dynamics can not only be used in enhancing data security, it can also be deployed in internet based surveillance applications, in which keystroke profile can be used to identify a remote user using the internet.

Today’s passwords are more complex than ever, but making them more and more complex has also resulted in user inconvenience and deteriorated login experience. When added with keystroke dynamics, passwords based security can improve multifold without introducing anymore complexity.

Pros

• Keystroke dynamics can improve password based security without introducing additional complexity.
• It works in the background and a user does not need to do anything else.
• It can be used in surveillance applications with keystroke profile of criminals or terrorists.

Cons

• Not suitable for identification or authentication applications like employee attendance, customer identification, door access etc.
• Implementation of keystroke biometrics may conflict with local laws in some jurisdictions.

Lip motion

In face to face conversation, lip motion helps us figure what other person is saying in hard-to-hear situations. Deaf people also make use of lip motion to comprehend others or to convey their message. Even in a normal human to human conversation, lip motion plays its part. Pattern of lip motion is considered to be a unique characteristic in human beings. As is the case with other biometric characteristics, there is no evidence or data to prove uniqueness of lip motion. However, we can understand by this: Just like gait, motion of lips depends on many factors like muscular movement, habitual factors, behavioral aspects, etc. Since these factors are bound to differ in each individual, their motion of lips follows a specific pattern and can be used to identify them with the help of biometric technology.

Image: Lip motion can verify if you your lip motion is as it should be while saying a password.

Unfortunately, not much work has been done in this direction and only isolated efforts exist. Cheung Yiu-ming, a computer science professor at the Hong Kong Baptist University leveraged lip motion biometrics to fix inadequacies associated with passwords as well as biometric methods. Passwords are insecure yet changeable if stolen, on the other hand, biometrics is secure but cannot be altered if compromised. Professor Cheung Yiu-ming thinks that lip motion biometrics could fix these shortcomings associated with passwords as well as classic biometric methods.

Passwords belong to a password prompt and are typed using a keyboard and keypad. We try to hide them while typing to save them from shoulder surfing. Even your password prompt obscures them with dots or asterisks so that a shoulder surfer cannot read them. You would never imagine typing a password with voice input. Lip motion password technology change that and wants you to say your password.

The lip motion recognition system not only require the password you said to be correct, but also verifies that your lips were moving correctly while saying it. Being an individual characteristic, your lips are expected follow the pattern they followed while setting the password. Lip motion recognition systems are designed to mathematically map movement of your lips and associate it with the password you set. Both are verified when you try to access your account.

Pros

• Lip motion biometrics can fix shortcomings of passwords by associating them with lip motion. Even if someone is able to guess or steal your password, it can be changed and the system can create a new lip motion profile.
• Lip motion biometrics also fixes shortcomings associated with classic biometric methods like fingerprint or iris recognition. Unlike other biometric identifiers like fingerprint or iris patterns, which cannot be changed if compromised, a new password with a new unique lip motion map can be created.
• Easy to setup on present day mobile and computing devices. Device camera is the only hardware required, rest of the system rests at the software side.

Cons

• Still in its infancy, long way to go mainstream.

Body odor recognition

Animals, specially mammals have extraordinary ability to recognize other animals, objects and humans with by their scent. Humans, despite being mammals, have very limited olfactory ability. However, researchers have been working to fill the gap with their efforts to give machines ability to recognize human beings with their body odor. Scientists at the Polytechnical University of Madrid worked on a biometric system that was able to recognize human being with their unique body odor. The research was accomplished by the Group of Biometrics, Biosignals and Security (GB2S) at the Universidad Politécnica de Madrid (UPM) in collaboration with Ilía Sistemas SL.

Human beings have ability to figure out minute differences visually; however, when it comes to do the same level of differentiation in smells, their ability fall short. Researchers found that each individual has a unique body odor that slightly differs from others, however, we it is hard to perceive it with our olfactory modality. The system designed by Polytechnical University researchers can sense this difference and identify an individual, whose identity has been pre-established for his or her unique body odor.

Pros

• It is extremely hard to replicate someone’s body odor, making the system immune against spoof or imposter attacks.

Cons

• Body odor recognition technology is still in its infancy and is yet to become commercially available.

Palm print recognition

Palm print recognition can be called the extended version of fingerprint recognition. The underlying technology used for scanning and processing the palm prints is fundamentally same as fingerprint recognition tech. However, the difference is that along with ridge information, palm prints also contain additional information like indents, texture, principle and secondary lines. Since human palm covers much larger area than fingertips, it is divided in different regions: 1. finger root region, 2. inside region and 3. outside region. Other information like geometry of palm, datum points, wrinkle features, delta point features and minutiae feature (as found in fingerprints) are also present in palm prints.

Palm prints have been used for identification for individuals at a crime scene in forensic and criminal investigations, their credibility in these applications has also paved way to commercial biometric applications. Automated palm print recognition systems make use of a digital camera or scanner to capture image of palm. Data acquisition may be contact or contactless type. This image is pre-processed and passed to the underlying recognition system, which does the job of feature extraction. A unique biometric template is generated and saved in the recognition system’s database, which is either used for enrolling or verifying the user identity.

Pros

• Since palm prints include way more details and distinctive features than fingerprints, they offer more security and make spoofing more difficult.
• The technology is fundamentally same as fingerprint recognition, it can be used along with fingerprints to take security even further.

Cons

• Since the recognition system has to capture larger skin area, equipments can be bulkier.

Palm vein recognition

Palm vein recognition systems use near-infrared light and a CCD (Charge Coupled Device) camera to capture the vein pattern. Near-infrared is used instead of visible light because deoxygenated haemoglobin in veins absorbs the near-infrared light. When near-infrared light source illuminates the palm under scanner, blood in veins absorbs it and veins appears as a black pattern on the image captured by the system’s camera. This image with palm vein pattern is pre-processed and enhanced by the system and finally the recognition system generates a unique biometric template, which can either be used to enroll an individual or to verify his/her pre-established identity.

Palm vein recognition is one of the biometric vein recognition methods that makes use of unique vein pattern found beneath the skin of palm of human hand. We have discussed about finger vein recognition above, and technology-wise, palm vein recognition is fundamentally same as finger vein recognition. The pattern formed by the veins beneath the surface of palm skin is considered to be unique for an individual, given the numerous possibilities of variations.

Pros

• Hidden beneath the palm skin, vein pattern is unexposed and can only be scanned with special setup, hence hard to replicate and circumvent the system.
• Being located beneath the skin, palm vein pattern is securer than fingerprints, face recognition and similar recognition methods in which biometric characteristics are exposed and can be collected without subject’s awareness.

Cons

• The technology is yet to mature and only a very limited numbers of applications are available for deployment.

Retinal scan

Blood capillaries appear like black lines on the resulting image. This raw image is pre-processed and enhanced to make it usable for the recognition system. The recognition system eventually generates a biometric template, which can be used for enrolling or verifying the subject.

Since retinal pattern is located deep in the eye, it cannot be acquired using ordinary methods like camera or any other superficial imaging. This pattern is scanned using special equipment called retina scanner, which require a subject to stare at the equipment optics steadily. Retina scanners scan pattern of retina blood capillaries using near infrared light. Near infrared is absorbed by the blood vessels and reflected by the surrounding tissue, which is captured by the inbuilt camera of the retina scanner.

Human eyes feature some characteristics that are considered to be unique for an individual and can be used to identify him or her. Pattern of blood vessels found on retina is one of those human eye characteristics. The posterior portion of human eye is called retina and it is made of tissue which can sense light. A thin layer of neural cells form retina and blood supply to this layer is carried out by vein capillaries. This network of blood capillaries forms a pattern on retina, which is considered unique for an individual.

Pros

• One of the most secure and extremely accurate methods of biometric recognition.
• Can be used where top level of security is required.

Cons

• Highly invasive, can cause user discomfort.
• Low collectability, special equipment is required for sample collection.
• High cost of implementation.
• One of the least deployed biometric identification methods due to cost and invasive nature.

Signature recognition

Signatures have been in use for personal identification in low as well as high value transactions. They have been used by banks and financial service providers as a mean of authentication and authorization. Signature is a behavioral characteristic and it can produce a lot of statistically significant data if captured electronically. While manual methods of signature verification include verification of its shape, biometric signature recognition system can verify a lot more to make sure that the signer if the authorized user and not an imposter.A biometric signature recognition system can verify the following information while capturing signatures:

• Spatial coordinate
• Azimuth
• Pressure
• Inclination

When captured biometrically, signatures are taken on a digitizing tablet to acquire data required for verification or enrollment. Biometric signature recognition system creates a profile specific to the signer and can enroll him/her on the system. Once enrolled and identity is established, the signer can be instantly verified whenever required in the future.

Pros

• Signature verification has been a reliable method of identity verification, inclusion of biometric technology improves its credibility.
• Biometrically powered signature verification improves the speed of verification in banking, finance, retail and similar applications.
• Biometric signature verification can be used in digital as well as paper based transactions.

Cons

• Since it is a behavioral biometrics that requires certain level of human ability, it excludes people who are illiterate and people who are not able to write their signature.

Image: Absorption of light spectrum is an individual characteristic and each individual’s skin absorbs and reflects it differently

Skin reflection

At superficial level, all skin types may look same except their color, however, there are minute differences that can be captured with the help of technology and can be used to uniquely identify an individual. Recognition of individuals on the basis of how their skin absorbs the light can be useful in many applications.

Lumidigm Inc., which is now part of HID Global, found that absorption of light spectrum is an individual characteristic and each individual’s skin absorbs it differently. Since absorption is different, the light spectrum is also reflected differently. In skin reflection based authentication setup, lights of different wavelength is sent into the skin and reflected light is read with photodiodes. Personal authentication based on skin reflection is in its experimental stage and there is no commercial offering are available as of now.

Pros

• Entire human body is covered with skin, hence a large area is available for authentication, unlike many other biometric recognition methods which depend on a small region (like fingerprints, iris, retina, etc.)

Cons

• Personal recognition on the basis of their unique skin reflection is in its very early stage and there are no commercial offerings as of yet.

Speaker recognition

Speaker recognition is a part of voice recognition that also includes speech recognition. Speaker recognition is the methodology of identifying a speaker, while speech recognition, as the name suggests, is identifying what is being said. Since biometrics is about identifying individuals with their behavioral or physiological characteristics, speaker recognition part of voice recognition relates with biometrics. Acoustic features of speech are an individual characteristic and can be used to identify a person on the basis of his/her pattern of voice. This pattern is called the voice print. Voice is considered to be a physiological as well as a behavioral characteristic in biometrics.

A speaker recognition system is used for the identification of an unknown identity or verification of an identity claim on the basis of a voice sample. The voice sample is converted into the voice print, which is either used to establish identity of an individual (enrollment) or to look for a match in the database of pre-established identities (verification).

Pros

• Non-invasive, easy to capture and setup.
• Ideal for use cases like phone banking, customer identification in call centers, etc., which results in saved time and cost in customer verification process.
• Speaker recognition can also be used for surveillance applications, in which it can help identify criminals and terrorists on the basis of their voice sample recorded using other surveillance techniques.

Cons

• Speaker recognition system may suffer with performance issues due to change in voice or user behavior due to disease, psychological condition, fatigue, aging, etc.
• Different equipment used by users (e.g. a different phone) can also have impact on recognition system’s performance.
• Highly vulnerable to imposter attacks as an unauthorised user may present recorded voice of an authorized user.

Image: Thermogram of a person with different temperature regions

Thermography recognition

Thermography recognition is a way of personal identification on the basis of thermograms. Thermograms are images or videos captured with thermographic cameras (also called a thermal imaging camera or infrared camera). While visible light cameras work in 400 – 700 nanometre range an infrared camera can capture 9 – 14 µm range of the electromagnetic spectrum. The infrared imaging leverages the fact that all objects (including human beings) above absolute temperature emit infrared radiation. When captured, these thermal images show different colors on the basis of temperature distribution. Pattern recognition methodology can make use of these thermograms and use them to uniquely identify an individual.

Thermography recognition primarily uses heat patterns of a person’s face. This heat pattern is greatly contributed by blood vessels located beneath the facial skin. Since pattern of blood vessels can differ in each individual, even twins are not found to have same facial heat map, as their pattern of blood vessels differ. It gives facial thermography an edge over facial recognition as facial recognition is dependent on external illumination with visible light, facial thermography recognition make use of infrared radiation inherently emitted by the face.

Pros

• Non-intrusive, non-invasive method of identification.
• Despite being fundamentally same as face recognition, facial thermography uses inherent infrared radiation and does not depend on external illumination.
• Can identify people from a distance so can also be deployed for security and surveillance.

Cons

• Higher cost of implementation than facial recognition.

Voice recognition

Voice recognition is a broad term that includes both speaker recognition and speech recognition. Speech recognition is a technology powered way to recognizing what is being spoken, while speaker recognition is about recognizing who is speaking, i.e. the identification of the speaker. Since speaker recognition and speech recognition have different objectives, they have entirely different approaches of implementation, except that both are related with human voice.

Speaker recognition technology leverages the fact that voice is an individual characteristics and everyone has a unique way of talking. Voice is a physiological as well as a behavioral characteristic. It depends on physical structure of throat and mouth as well as habitual factors. Being dependent on many factors, voice becomes a biometric identifier which can be used to identify the speaker. A person’s spectrogram or voice print is used for the purpose. A visual record of speech, analysed with respect to frequency, duration, and amplitude is called the voice print or spectrogram.

Speech recognition, on the other hand, is about what is being said. Recognition and translation of what is being said by a user has become more apparent these days due to rise of smart devices and digital assistants, You can communicate what you want and the system will execute the appropriate action, e.g. turning on the light by saying “turn on the lights” in a smart home system.

Traditionally, approaches like Hidden Markov Model have dominated the implementation speech recognition methods; however, modern systems make use of Artificial Intelligence, Neural Networks, Big Data and Deep Learning methods like Long short-term memory (LSTM) to implement speech recognition.

Pros

• Voice, being a natural way to communicate, can fill the gap between human and machine.
• Speaker identification systems can be deployed in call centers and phone banking systems to recognizing callers without wasting time in identification formalities.

Cons

• Prone to spoofing attacks with recorded voice of an authorized user.
• May not be suitable for high security access control.

Conclusions

That is not all. We have covered many biometrics above but there are more physiological and behavioral characteristics that can be used to identify human beings uniquely. Their implementation, however, is mostly limited by technology or lack of their commercial significance. With the rise of AI and machine learning, biometric systems are also set to make use of it and improve over time. The day is not far when there will absolutely zero interference of biometric system and people will be in the state of “continuous authentication”.