Biometrics verifies individuals based on who they are i.e. the characteristics inherent to the individuals. These characteristics (also known as modalities) uniquely identify individuals from an entire population based on their intrinsic physical or behavioral traits.
These modalities have various key advantages such as non-repudiation, not transferable, not guessable and also provide a very high level of protection against fraud. The technology have been successfully implemented in various real-life applications such as forensics, government agencies, banking and financial institutions, enterprise identity management and other identification and recognition purposes.
Each of the different modalities has something to recommend them. For example, some traits are less invasive and can be done without the knowledge of the subject while others are very difficult to fake. We will discuss and compare the top five biometric modalities.
Variation in quality – biometric accuracy comparison
Biometric systems have been utilized in many large-scale deployments all over the world. Some prominent examples include the UK Iris Recognition Immigration System (IRIS) project, US Visitor and Immigration Status Indicator Technology (VISIT) and India’s Aadhar project. These far-reaching projects have not only assisted and enhanced our current way of life but have also opened up new research directions. However, there exists a very important research challenge and that is the measurement of quality of a sample. Similar to other pattern recognition and machine learning applications, these systems also get affected by the quality of input data. This makes it extremely important to evaluate the quality of a sample as it provides an indication of its ability to function as a biometric.
For example, quality measures are very important in facial recognition systems as large degrees of variations are possible in face images. Factors such as illumination, expression, pose and noise during face capture can affect the performance of facial recognition systems. The performance of iris based systems is also highly dependent on the quality of the captured sample. Iris recognition covariates include focus and motion blur, dilation or constriction, cosmetic contact lenses, spectacles etc. The performance of samples that are captured in indoor and studio-like conditions will be better than the performance of samples captured in outdoor and uncontrolled conditions.
Therefore, it is essential to perform quality assessment and feedback during verification to mitigate any false matches. If the quality score is below a particular threshold, the verification system may choose not to perform the matching. Nowadays, many fingerprint and iris sensors come with active quality-control mechanisms.
What is the most accurate biometric modality?
Although many modalities are available, it is important to realize that not all traits have the ability to meet the requirements of every organization or application and it depends to a large extent on the industry application context. Hence, there are several factors which need to be carefully taken into account before selecting a particular modality. These factors include the unique requirements, environment, and circumstances of the application as well as the acceptability and ease of use. It should however be noted that performance and cost-effectiveness will vary significantly depending on the deployment requirements and environment. Environment may include factors such as working conditions of the people who will be using the system on a regular basis, the way they will use it and feasibility. Some deployments may also require combining two traits known as multimodal biometrics to ensure optimal accuracy. The selection of the right modality is extremely important to reap the full benefits of a biometric system as well as to ensure maximum return on investment.
It is one of the most important factors that need to be assessed when selecting a modality. Again, accuracy is based on several other factors such as false acceptance rate (FAR), false reject rate (FRR), error rate, identification rate etc.
The widespread use of biometric recognition systems in various sensitive applications stresses the importance of stronger protection against intruder attacks. Therefore, a lot of importance is given to direct attacks where unauthorized individuals can gain access to the system by interacting with the system input device. These unauthorized attempts to access the system are known as spoofing attacks and therefore the chosen modality should have strong anti-spoofing capabilities.
This is an important factor to consider when deciding the effectiveness and suitability of a particular modality. Some modalities may be more cost-effective than others due to the underlying technology or hardware characteristics. It is important to realize that the initial investment done on a biometric system can often be compensated in a short amount of time which leads to faster return on investment (ROI).
The deployment of a particular identification system also depends on how well it is accepted by the users. In some cultures, certain modalities have a stigma associated with them and it can negatively impact the success of the implemented modality. Therefore, it is important to understand which modalities are well acceptable versus those that may cause some user acceptance issues.
Another important factor to consider before making a deployment decision is whether the system has contact dependent hardware. Many organizations prefer to use contactless modalities due to hygiene reasons and also for infection control.
So organizations should consider all the above factors before selecting a particular modality for their applications. The selected modality should also meet the operational requirements for their deployment.
What is the best biometric? – Our verdicts and recommendations
Biometric systems utilize various physical or behavioural characteristics for identifying individuals and every trait has its own strengths and weaknesses. The top five modalities are discussed in the following sections that will help you to decide the right modality for your application environment.
It is one of the most flexible biometric identification methods which can work even when the subject is unaware of being scanned. This method shows a very promising way to search efficiently through masses of people who spend only few seconds in front of a scanner i.e. a digital camera. The working of facial recognition system is based on analyzing certain features that are common to every individual’s face. These features include the distance between the eyes, position of cheekbones, jaw line, chin, width of nose, shape of mouth etc. Every individual can be uniquely identified by combining these numerical quantities into a single code.
This system can automatically identify or verify an individual from a digital image or a video frame. It achieves this by comparing the selected facial features from the image to a facial database and is typically used in security systems.
It has a medium accuracy rate. There are several factors that can affect the accuracy of facial biometrics. It might not work well under poor lighting conditions, the presence of sunglasses or other objects that partially cover the subject’s face and low resolution images. Also the face of a person changes over time. The accuracy of some facial recognition systems is also affected by the variation in facial expressions and for this reason most countries allow only neutral facial expressions in passport photos.
In automated border control gates at immigration, a live photo is captured of the passenger. Facial recognition technology is then used to compare this live capture with the photograph read off the passenger’s passport chip. The passenger is allowed to proceed on a positive match else the passenger is assisted by a border security guard. In this case, FAR represents the percentage of travelers holding a passport that does not belong to them and are wrongly admitted. FRR represents the percentage of legitimate passengers who are wrongly directed to a border security guard due to mismatch of photographs.
A real-world face biometrics accuracy test verified passport photos of passengers against a lesser quality live scan photo taken within the border control gates itself. The top performing vendor in the National Institute of Standards Technology (NIST) test achieved a FRR of 1.1%. Thus, face recognition systems definitely provide better accuracy when compared to live guards performing a manual comparison of passport photos with the passport holders.
Face as a biometric trait is gaining increasing acceptance and is now one of the most commonly used credentials for identifying individuals. It is used in applications such as border control and immigration, government initiatives such as national ID, voter registration, passport and also for large workforce management. Although it may not have a very high rate of accuracy and reliability, one primary advantage of this system is that it does not need the co-operation of the test subject to work.
It is hands-free and non-intrusive identification method. It is highly approved by a majority of users as the identification can be performed from a distance. It can be utilized in static applications such as mug shots and dynamic applications such as airports.
Fingerprint based security systems are widely popular due to their potential to reliably identify people based on a near-universal physical trait. Moreover, the advancements of technology have led to development of fingerprint recognition systems that are small and inexpensive. As a result, the deployment of these systems has increased in a wide range of situations and applications. Some prominent examples of its application include mobile phones and laptops, building and car doors, border control and other high security military applications.
Fingerprints are typically made up of ridges and furrows and its uniqueness is determined by the patterns made by the ridges, minutiae points and furrows. The finger ridge configurations remain unchanged throughout the life on an individual and most importantly even twins have different sets of fingerprints. This characteristic makes it a very popular identifier and therefore it has been used in personal identification systems for a very long time. It is the most widely used modality owing to its distinctiveness and stability.
These systems have a very high rate of accuracy. Fingerprint patterns are so unique that it can be used to distinguish individuals from en entire global population. It is also extremely difficult to spoof under normal conditions. Various tests conducted for determining fingerprint accuracy suggests that its performance is extremely good for high quality images and when done under controlled conditions.
Tests conducted by the International Biometric Group on fingerprint systems of participating vendors found that the false acceptance rate of these systems ranged from 0% to 5%. On the same day of enrolment, tests were conducted for false rejection rate and it ranged from 0% to 35%. However, when the tests were conducted six weeks later the FRR’s ranged from 0% to 66%. Systems from some vendors worked very well while others had some accuracy problems. When vendor-independent tests were conducted by the FVG2006, it found that FAR was held constant at a rate of 0.01%. This rate is considered sufficient for most authentication scenarios.
For 4-print fingerprint: FAR=0.01% and FRR=0.1%
It is the oldest and the most deployed biometric technique in various industries. Of all traits, fingerprints have an extremely high rate of accuracy and affordability due to its uniqueness and the wide availability of low-cost sensors. Many security systems such as access control and encryption services incorporate fingerprint devices that enable them to provide a seamless environment.
Fingerprints are a near-universal biometric trait and this technology has been successfully used in various applications for more than a century. Its popularity can also be attributed to its ease of acquisition and numerous sources for collection (ten fingers). It can be easily integrated into existing applications with very low learning curve. Users can easily learn to use these systems as it is intuitive and needs no special training.
Various factors have helped to increase the usability of this technology. These factors include the availability of smaller, cheaper and more reliable sensors. Moreover, sensors are now designed with better ergonomic characteristics and come embedded in many consumer devices such as keyboards, cell phones and mice. Biometric algorithms are also improving and many fingerprint systems now include features that train users and provide feedback during usage.
A method of biometric authentication that uses pattern recognition techniques based on high-resolution images of the irises of an individual’s eye. Iris recognition utilizes the same technology as cameras coupled with subtle infrared illumination to reduce the specular reflection from the convex cornea. This creates detail-rich images of the intricate structures of the iris. When these images are converted and stored as digital templates, they provide mathematical representation of the iris that enables unambiguous positive identification of individuals.
Iris is formed in the early stages of an individual’s life and once it is fully formed its texture remains stable throughout a person’s life. The iris of the eye has a distinct pattern and iris recognition has been found to be a highly accurate biometric system. Its efficiency is rarely impeded by the presence of glasses or contact lenses. Moreover, it has a small template size that allows speedy comparisons making iris recognition technology particularly well suited for one-to-many identifications. Even genetically identical individuals have distinct iris textures which further confirm that it is a highly accurate and reliable technique.
National institute of standards technology conducts many evaluation tests to determine accuracy of biometric devices and the accuracy proven by NIST tests is considered to be the international recognition for the these devices and superior technology provided by vendors. The test for iris recognition systems conducted by NIST where FAR and FRR are the evaluation metrics is known as Iris Exchange or IREX test.
Iris recognition is a very stable technique with high template longevity where a single enrollment can last a lifetime. Since iris is an internal organ, it is very well-protected against damage and wear by a transparent and sensitive membrane known as the cornea. This feature distinguishes irises from fingerprints which can be quite difficult to recognize after certain years of manual labour. Also, the geometric configuration of the iris is only controlled by two complementary muscles. This makes the shape of the iris far more predictable than that of the face. However, iris scanners are relatively expensive as compared to other modalities and require user-cooperation. Iris recognition systems have been implemented in all of the UAE’s air, land and sea ports of entry. Google too uses this technology to regulate access to its datacenters. The FBI has also incorporated it into its next-generation biometric identification system.
Although iris recognition is the most accurate biometric system and works very well for positive identification against a large database, there are some usability concerns. It is a new technology that requires substantial investment and hence may not be suitable for small organizations. It is quite difficult to perform iris recognition from a distance larger than a few metres and moreover the subject to be identified needs to be co-operative. The subject should hold his or her head still and look into the camera. Iris recognition is also susceptible to poor quality of images as well as associated failure to enroll rates. However, iris has a fine texture similar to that of fingerprints and is formed randomly during embryonic gestation. This fine texture remains stable for many decades and attributes iris recognition to be the most accurate modality. Some iris identification schemes have succeeded over a period of almost 30 years.
Palm vein recognition
Palm vein is one of the most secure biometrics and is the world’s first contact less personal identification system. It works by capturing the vein pattern image of an individual while radiating it with near-infrared rays. Its specialty is that it can detect the vein pattern on the human palm with utmost precision. When the sensor emits a near-infrared ray towards the palm of the hand, the blood flowing through these back to the heart with reduced oxygen absorbs the radiation and causes the veins to appear as a black pattern. This pattern is then recorded and stored in an encrypted format in a database, token or smart card as a reference for future comparison.
Each time a person wants to log in by using his palm scan to bank account or other device, the newly captured image is also processed likewise and then compared to the stored one for verification purpose. The number and positions of veins along with their crossing points are compared and depending on the verification results the person is either granted or denied access.
It uses the information that is contained within a person’s body to confirm his or her identity. Therefore it is highly accurate because the vein pattern of the human palm is not only complex but it is also unique to each individual. It uses a combination of image recognition and optical technology to scan the normally invisible vein pattern of the human palm. This makes it highly resistant to any spoofing methods such as impersonation, counterfeiting or any other dishonest actions.
Correct recognition rate and verification rate can be used to evaluate an automated palm vein system. Verification rate is calculated using false acceptance rate, false rejection rate and equal error rate. FAR denotes the percentage of claims that are accepted but not genuine over the total number of not genuine accesses. FRR denotes the percentage of genuine claims that have been rejected over the total number of genuine accesses.
Veins are internal to the body and have a wealth of distinguishing features. It is extremely difficult to fool this technology by assuming false identity through forgery and therefore palm vein is highly secure. Moreover, it is designed in such a way that it can detect only the vein pattern of living persons. It also has a fast scanning process and does not need any contact which means that this technology meets the strict hygiene requirements that are usually required for usage in public environments.
This technology has the potential to be applied in a wide range of sectors such as banking, healthcare, educational facilities and commercial enterprises. Its applications include allowing physical access for secured areas, logging into computers or server systems, accessing POS, ATMs. Typically, the palm vein device is a small scanner that is simple and natural to use. It is also fast and highly accurate. Its contact less feature also gives it an added and hygienic advantage over other biometric technologies.
Voice recognition systems perform the task of validating an individual’s claimed identity by using certain characteristics extracted from his or her voice. A typical example of voice verification is in telephone banking systems where the system operates with the user’s knowledge and also requires co-operation from the users. It is also possible to implement these identification systems covertly without the user’s knowledge. For example, it can be used to identify the speakers in a discussion or alert automated systems of speaker changes. Forensic applications commonly perform a speaker identification process initially to create a list of best matches. Then a series of verification processes are done to determine a conclusive match.
In these systems, the emphasis is on the vocal features that produce speech and not on the sound or pronunciation of speech. Various attributes such as dimensions of the vocal tract, mouth and also other speech processing mechanisms of the human body can affect the vocal properties.
Automated voice recognition systems i.e. computers can better analyse and identify voice than people. Sometimes people might not know who is on the other end of a phone call or even when calling someone who they know well. However, such systems are able to accurately identify individuals from their voices with less than one percent error rate. The error rate is even lower for speakers that are saying a pre-determined phrase. The accuracy of these systems is almost similar to fingerprint systems.
Accuracy is defined through a combination of false positives and missed identifications and depends on various factors. These factors include the analysis and the voice print technology that is used, the length and quality of the audio recording that is being analysed and also if the analysis includes a pre-determined passphrase or sentence.
This technology can be evaluated on various techniques such as false acceptance rate and false rejection rate. For sensor subject distance of 20 cm, false acceptance rate is 2% and false rejection rate is 10%.
Speech recognition systems are typically used in situations where the only available biometric identifier is voice. These systems are inexpensive and require very less investment in hardware as most personal computers already have an inbuilt microphone. Voice verification systems are quite reliable, easy to use as well as no special instructions are needed.
Voice based systems usually require co-operation from the subject. In such systems, feeding the wrong voice always cannot be avoided and also the voice capturing machine should be near to the subject. The voice quality gets affected by noisy environments and also depends on the emotional condition of the users. A person’s voice may also change due to some health conditions such as cold etc. which might affect recognition. The size of the voice template database is large and it can impact the matching speed.