With the emergence of the internet, some talking points were raised, such as data security. In a scenario where over 80% of the population is connected, as it happens in Brazil, new technologies are created for safer navigation. Some are already popular, such as biometrics.
Biometric security and identification
When authenticating a user, it is important to ensure that access made through biometrics does not fail. A very common example of using this powerful identification tool is registration and access to a digital bank account.
A survey conducted by digital bank N26, in partnership with Accenture, showed that Brazil has the second highest growth in user support to digital banks in the world (with a 78% rate of participation), behind only Switzerland (with 82%).
In addition to that, this same study stated that the Brazilian population is in third place in the largest number of customers with digital accounts worldwide.
These numbers show the great popularity of digital banks, and them, aware of their customers' concerns, use biometric identification as one of their resources to ensure security.
Since the subject is delicate, since the security of our financial lives comes on play, all protection is welcome.
Noting the popularity of biometric identification in our daily lives, the question arises: can we do something to improve our security when we do biometric operations?
Yes! Thanks to the qualification of image in biometric correspondences.
What is biometric correspondence?
Biometric correspondence is based on the concept that we all have unique trace combinations that differentiates us from other people.
Thus, when capturing a face photo, for example, it is possible to identify a person if the biometric data contained in the photo corresponds to the previously collected data.
Although this analysis can be done by a human, nowadays the fastest and most efficient way to perform these operations is to use a machine that can, in a matter of seconds, verify the image and determine the identity of an individual.
And that's where image qualification comes into play —a good-quality image that highlights the individual's traits, allows for quick reading and improves not only the speed that the machine will perform these operations, but also decreases the rate of errors that eventually may occur.
As such, how can we do to qualify what is or not a good biometric capture?
Understand image qualification
To answer this question, the ISO/IEC 19794-5 standard was created, elaborated, as the name suggests, by ISO and IEC (organizations specialized in international standardization), whose objective is to describe the best way to collect biometric face data.
This standard is also called the ICAO standard, since it is used by the International Civil Aviation Organization to standardize the photos used in individuals' passports, reinforcing the global relevance of the use of biometrics as a protective factor.
In addition to the security already mentioned, the standard has as main objectives:
- Facilitate the analysis of facial identification performed by machines.
- Increase the accuracy of facial analysis operations.
- Allow biometric applications to perform efficiently even on machines with few computational resources.
To achieve this goal, the standard regularizes elements of the capture, such as: facial expressions, face position, ambient lighting, positioning and focus of the camera, size and resolution of the image, among others.
Based on this standard, some tips for performing a good capture of facial biometrics are:
- Avoid facial expressions (such as smiling or frowning).
- Keep your mouth closed.
- Avoid wearing accessories that cover your face (such as glasses or caps).
- Ensure to be in a well-lit environment, with the light aimed directly at yourself.
- Make sure to take a capture in an environment with a uniform background.
Although these are good measures and, in general, a good idea to follow them, it is not necessary human monitoring to ensure compliance with this standard.
Just as a machine can perform biometric capture and matching operations, this verification can be done through automatic operations guided by artificial intelligence.
AI in image qualification
An example of an application that helps qualify images for facial biometrics is the Quality API.
Through state-of-the-art artificial intelligence, this tool analyzes facial captures and points out which of the points of the strict standard proposed by ISO/IEC are being followed, ensuring more efficiency in your biometric operations.
In addition, the Quality API has the possibility to also qualify fingerprint captures, following the NFIQ standard, which, just like ICAO, was created to standardize the qualification of fingerprint images.
Although biometric operations work without the need to perform quality verification in advance, this type of analysis ensures more efficiency in biometric operations and reduces the error rates of your biometric system, quickly and with easy implementation.
