UIDAI: Banking on biometrics to give you a unique identity
Shelley Singh, ET Bureau Apr 26, 2012, 02.42AM IST
At the end of its enrolment drive, the Unique
Identification Authority of India (UIDAI) will collect about 12
billion fingerprints and 2.4 billion irises, besides 1.2 billion
photographs - the largest human biometric data ever collected.
At almost a quarter of the population already enrolled for Aadhaar and the unique identity project surpasses the US human database. Aadhaar also promises to be the panacea for e-governance in India - no duplicates, no frauds and one that an individual need not remember, like a password.
As pilots --using Aadhaar to authenticate people -- convert to actual rollout, people will be able to use fingerprints to avail of government subsidies, insurance policies, buy fertiliser or open bank accounts.
Behind this mega project and an aspect responsible for its most essential feature - uniqueness that can't be duplicated - lies biometrics or mathematical calculation of a human feature - eyes, face, palm, toes, fingers, veins and so on. Fingerprints change the least over time and are relied upon the most for biometric authentication followed by iris (see graphics -- Basics of Biometrics).
When UIDAI spearheaded by technology titan Nandan Nilekani was entrusted with the task of giving each Indian an unique identity that will transform delivery of services, this new e-governance initiative fell upon an ancient idea, biometrics -- specifically fingerprints and iris -- to create uniqueness.
Biometrics have been around since 29000 BC when cavemen would sign their drawings with handprints. In 500BC, Babylonian business transactions were signed in clay tablets with fingerprints.
The earliest cataloguing of fingerprints dates back to 1881 when Juan Vucetich, an anthropologist and police officer who started collecting fingerprints of criminals in Argentina. All governments, notably the police departments, use fingerprints and in more than 100 years of database available, no two fingerprints, have been found to be identical. Hence a fingerprint uniquely identifies an individual.
In the case of Aadhaar, biometrics will also translate into a huge business for iris and fingerprint device makers, besides network services providers. Each iris scanner costs about Rs 10,000 and fingerprint scanner almost double that rate.
Of Ridges & Valleys
Fingerprint comprises 'ridges' and 'valleys' which form the basis for the loops, arches, and contours seen on fingertips. The ridges and valleys have different kinds of breaks and discontinuities. These are called 'minutiae'. From these minutiae, the unique features are located and determined.
Each finger has four to half-a-dozen minutiae points and there are in turn, two types of minutiae: ridge endings -- the location where the ridge actually ends -- and bifurcations, the location where a single ridge becomes two ridges.
Based upon the unique features found in the minutiae - the location, position, as well as the type and quality of the minutiae -a `template' is created. Template is smaller in size than a raw image and is based on ISO 197942-2 standard.
This standard specifies formats on how minutiae points are determined and collected. The templates thus formed are saved into a database, which in case of Aadhaar, is located at the Bangalore technology hub of UIDAI. To use fingerprints for authentication, Aadhaar uses algorithms certified by the US-based National Institute of Standards & Technology (NIST).
Says Ankur Saluja, co-founder SmartID, "The thumb and the next two fingers (index and middle) have the maximum information and best suited for fingerprint identification." SmartID, a 2008 start-up which has partnered with the US-based Cross Match Technologies to provide biometric solutions in India.
The other biometric, iris essentially implies capturing the pattern of the iris. Iris recognition uses camera technology with subtle infrared illumination to acquire images of the detail-rich, intricate structures of the iris. Digital templates encoded from these patterns by mathematical and statistical algorithms allow unique identification of an individual.
Both irises are captured to form an iris template, which ranges from type 1 to 7-capturing high-resolution images of irides of the eye.
Unique, But Can't Authenticate
While biometric twins have not been found so far, making them the best means to identify and authenticate individuals, field trials of Aadhaar have revealed gaps in authentication. All the people given Aadhaar numbers could not be authenticated by merely putting a finger on a scanner. All scanned prints did not match with ones stored in Bangalore database.
For instance in the first attempt to authenticate, a high 40% of the farm workers, could not be verified though this improved to more than 90% in three attempts -- one person trying different fingers to authenticate. That's because while fingerprints are unique, authentication accuracy depends on multiple of factors, ranging from cleanliness of fingers to size of fingerprint scanners, leading to mixed results.
Explains
Anil K Jain, professor in the department of computer science at the Michigan
State University and a holder of six patents on fingerprinting
technology: the first is condition of the finger itself. That is whether
the finger have cuts, abrasions, wet, too dry, dirty, and so on.
Second, he says, is the quality of the fingerprint image acquired by the fingerprint scanner. The quality of the acquired image will depend on the first factor, namely the condition of the finger, but it also depends on how the user is interacting with the sensor (finger placement on the sensor and the applied pressure) and the condition of the sensor platen or surface (could be dirty), and third, robustness of the fingerprint matcher.
Says Jain: "Given these confounding factors, we cannot guarantee that fingerprint authentication will be 100% accurate. However, for a reasonable quality fingerprint image and a high grade fingerprint matcher, we can be assured that authentication accuracy will be very close to 100% (say, over 99%). Field deployments and testing support this."
Age also play a role in fingerprint matching. Says SmartID's Saluja, "after age 60, fingerprints are too dry. Moist fingers reveal better patterns. And a 15 year old, will need to give fresh prints at age 40 to ensure scanners don't reject him."
To ensure higher accuracy, wherever biometrics is needed, UIDAI is now looking at iris as well. To begin with iris was being used only to ensure there are no duplicates for the Aadhaar numbers.
Says Saluja,"Iris is more stable and remains unchanged after two years of age, compared to other biometrics, including fingerprints that can change over time." Perhaps a combination of two, will work best to ensure each of the 1.2 billion individuals is unique.
Comments:
At almost a quarter of the population already enrolled for Aadhaar and the unique identity project surpasses the US human database. Aadhaar also promises to be the panacea for e-governance in India - no duplicates, no frauds and one that an individual need not remember, like a password.
As pilots --using Aadhaar to authenticate people -- convert to actual rollout, people will be able to use fingerprints to avail of government subsidies, insurance policies, buy fertiliser or open bank accounts.
Behind this mega project and an aspect responsible for its most essential feature - uniqueness that can't be duplicated - lies biometrics or mathematical calculation of a human feature - eyes, face, palm, toes, fingers, veins and so on. Fingerprints change the least over time and are relied upon the most for biometric authentication followed by iris (see graphics -- Basics of Biometrics).
When UIDAI spearheaded by technology titan Nandan Nilekani was entrusted with the task of giving each Indian an unique identity that will transform delivery of services, this new e-governance initiative fell upon an ancient idea, biometrics -- specifically fingerprints and iris -- to create uniqueness.
Biometrics have been around since 29000 BC when cavemen would sign their drawings with handprints. In 500BC, Babylonian business transactions were signed in clay tablets with fingerprints.
The earliest cataloguing of fingerprints dates back to 1881 when Juan Vucetich, an anthropologist and police officer who started collecting fingerprints of criminals in Argentina. All governments, notably the police departments, use fingerprints and in more than 100 years of database available, no two fingerprints, have been found to be identical. Hence a fingerprint uniquely identifies an individual.
In the case of Aadhaar, biometrics will also translate into a huge business for iris and fingerprint device makers, besides network services providers. Each iris scanner costs about Rs 10,000 and fingerprint scanner almost double that rate.
Of Ridges & Valleys
Fingerprint comprises 'ridges' and 'valleys' which form the basis for the loops, arches, and contours seen on fingertips. The ridges and valleys have different kinds of breaks and discontinuities. These are called 'minutiae'. From these minutiae, the unique features are located and determined.
Each finger has four to half-a-dozen minutiae points and there are in turn, two types of minutiae: ridge endings -- the location where the ridge actually ends -- and bifurcations, the location where a single ridge becomes two ridges.
Based upon the unique features found in the minutiae - the location, position, as well as the type and quality of the minutiae -a `template' is created. Template is smaller in size than a raw image and is based on ISO 197942-2 standard.
This standard specifies formats on how minutiae points are determined and collected. The templates thus formed are saved into a database, which in case of Aadhaar, is located at the Bangalore technology hub of UIDAI. To use fingerprints for authentication, Aadhaar uses algorithms certified by the US-based National Institute of Standards & Technology (NIST).
Says Ankur Saluja, co-founder SmartID, "The thumb and the next two fingers (index and middle) have the maximum information and best suited for fingerprint identification." SmartID, a 2008 start-up which has partnered with the US-based Cross Match Technologies to provide biometric solutions in India.
The other biometric, iris essentially implies capturing the pattern of the iris. Iris recognition uses camera technology with subtle infrared illumination to acquire images of the detail-rich, intricate structures of the iris. Digital templates encoded from these patterns by mathematical and statistical algorithms allow unique identification of an individual.
Both irises are captured to form an iris template, which ranges from type 1 to 7-capturing high-resolution images of irides of the eye.
Unique, But Can't Authenticate
While biometric twins have not been found so far, making them the best means to identify and authenticate individuals, field trials of Aadhaar have revealed gaps in authentication. All the people given Aadhaar numbers could not be authenticated by merely putting a finger on a scanner. All scanned prints did not match with ones stored in Bangalore database.
For instance in the first attempt to authenticate, a high 40% of the farm workers, could not be verified though this improved to more than 90% in three attempts -- one person trying different fingers to authenticate. That's because while fingerprints are unique, authentication accuracy depends on multiple of factors, ranging from cleanliness of fingers to size of fingerprint scanners, leading to mixed results.
Second, he says, is the quality of the fingerprint image acquired by the fingerprint scanner. The quality of the acquired image will depend on the first factor, namely the condition of the finger, but it also depends on how the user is interacting with the sensor (finger placement on the sensor and the applied pressure) and the condition of the sensor platen or surface (could be dirty), and third, robustness of the fingerprint matcher.
Says Jain: "Given these confounding factors, we cannot guarantee that fingerprint authentication will be 100% accurate. However, for a reasonable quality fingerprint image and a high grade fingerprint matcher, we can be assured that authentication accuracy will be very close to 100% (say, over 99%). Field deployments and testing support this."
Age also play a role in fingerprint matching. Says SmartID's Saluja, "after age 60, fingerprints are too dry. Moist fingers reveal better patterns. And a 15 year old, will need to give fresh prints at age 40 to ensure scanners don't reject him."
To ensure higher accuracy, wherever biometrics is needed, UIDAI is now looking at iris as well. To begin with iris was being used only to ensure there are no duplicates for the Aadhaar numbers.
Says Saluja,"Iris is more stable and remains unchanged after two years of age, compared to other biometrics, including fingerprints that can change over time." Perhaps a combination of two, will work best to ensure each of the 1.2 billion individuals is unique.
Comments:
Karthik (Chennai)
26 Apr, 2012 01:09 PM
I believe that fingerprint does not change with age. We have to use the right technology to enroll. Rural India, most of the fingerprint are not clear. I have studied that each finger has three layers which is exactly the same. Radio frequency technology will exactly solve the problem for enrollment and authenticate. RF Scanners takes the inner layer i.e. Corium layer of the finger to enroll and authenticate. RF scanners are successful in financial inclusion which again is implemented only for rural India. I believe that we have to prefer RF technology for enroll and authenticate, as Optical imaging technology is failing everywhere.
26 Apr, 2012 01:09 PM
I believe that fingerprint does not change with age. We have to use the right technology to enroll. Rural India, most of the fingerprint are not clear. I have studied that each finger has three layers which is exactly the same. Radio frequency technology will exactly solve the problem for enrollment and authenticate. RF Scanners takes the inner layer i.e. Corium layer of the finger to enroll and authenticate. RF scanners are successful in financial inclusion which again is implemented only for rural India. I believe that we have to prefer RF technology for enroll and authenticate, as Optical imaging technology is failing everywhere.
Vickram
(Mumbai)
26 Apr, 2012 08:54 AM
The article does not comment on the fact that fingerprints often change with age, more so with manual labor, making this method even more expensive to deploy for authentication in rural India, said to be the principal target for beneficiaries of social solutions employing Aadhaar. Of course, most of your readers will already be aware that 1% of the population exceeds 12 Lac people who will face unprecedentedly huge problems - exclusion from banking and receipt of subsidies - in establishing their identities if UIDAI fails them.
26 Apr, 2012 08:54 AM
The article does not comment on the fact that fingerprints often change with age, more so with manual labor, making this method even more expensive to deploy for authentication in rural India, said to be the principal target for beneficiaries of social solutions employing Aadhaar. Of course, most of your readers will already be aware that 1% of the population exceeds 12 Lac people who will face unprecedentedly huge problems - exclusion from banking and receipt of subsidies - in establishing their identities if UIDAI fails them.
26 Apr, 2012 12:53 PM
I believe fingerprint does not change with age. Each finger is unique and permanent. We are not able to capture the fingerprint is because we are using optical imaging technology. Each finger has three layers, so if we can capture the inner layer, which is called the corium (inner layer) layer, then it is easy to authenticate any finger. Radio frequency based finger print scanners is able to do this.
I believe fingerprint does not change with age. Each finger is unique and permanent. We are not able to capture the fingerprint is because we are using optical imaging technology. Each finger has three layers, so if we can capture the inner layer, which is called the corium (inner layer) layer, then it is easy to authenticate any finger. Radio frequency based finger print scanners is able to do this.
