Intel RealSense eases 3D facial authentication integration for diverse use cases

facial-recognition-database

Rapid adoption of facial authentication is a major trend across a wide variety of industries, as businesses carry out digital transformation and move away from legacy identity management technologies to increase their security, improve user experiences and enable new capabilities.

Active Stereo technology enables highly accurate 3D authentication and liveness detection to deliver a higher level of security than legacy systems used in physical or logical access control. With Intel® RealSense™ ID Solutions, this technology can now be built into a nearly endless variety of applications, including as part of smart locks, gate access control, kiosks, ATMs, and point-of-sale systems.

Facial authentication gives physical access control systems like smart locks or gates high security without requiring physical contact, and often even without an ID card or badge. For kiosks, face authentication can enable fast, low friction processes, while for PoS systems, face authentication holds the promise of one-step, touchless payment authorization without requiring the customer to use any personal device.

Moving face biometrics to the network edge addresses a range of concerns and confers several advantages. Flexibility of deployment, however, is more important for the many businesses, including those for whom a centralized server or cloud-based architecture is better suited.

Integrators are also beginning to develop solutions that apply the power of 3D facial authentication to the needs of their customers and partners. As long as the system architecture is effective, many businesses of various sizes, in different verticals, can achieve significant operational and efficiency improvements with 3D facial authentication.

Supporting the challenges of a diverse range of businesses

For small and medium sized businesses looking to implement 3D face authentication, an edge architecture can be an easy way to upgrade while taking advantage of existing resources.

The database can be stored on the capture device or the host controller it is connected to, either of which is sometimes referred to as the network ‘edge.’ It can also be stored on a central server, in the cloud, or in multiple places.

For businesses with larger workforces, a database hosted on a local server can utilize an existing database, and store tens of thousands of users or more.

For enterprises or businesses that need to authenticate distributed workforces or customer bases, a cloud architecture enables authentication on any terminal or endpoint in any location within the network.

Where the database resides and the biometric matching algorithm runs determines the relative strengths and weaknesses of each architecture. These range from inexpensive, nimble systems running mostly on legacy technology for smaller businesses to vast, cloud-enabled implementations that can serve the needs of globally dispersed workforces of customer bases, still with robust security and a smooth user experience.

Where is the database?

The key question to answer for a business to get the most out of a new face authentication system, therefore, is where the database of biometric information will reside. That decision must be made according to the number of people using the system, where they will be when they use it, the performance requirements of the system itself, and even regulatory compliance requirements.

The different locations within a network where the database can be stored are listed above, but getting the most out of face authentication technology investments means carefully selecting between a range of architectural possibilities.

Edge architecture can take the form of data storage and matching on the device itself, on a local server, or a combination of the two. A ‘hybrid’ edge architecture refers to implementations in which the matching may be performed locally or in the cloud. This can be advantageous for organizations with different databases, such as one for employees that functions as part of a workforce management system, and another for customers who may visit different locations.

Processing at the edge also delivers results without latency, reduces resource use, and can help protect user privacy.

In a cloud architecture, the database of biometric templates is stored remotely, and images are also compared by the matching algorithm in a cloud environment.

In some cases, organizations will want to perform 3D face authentication using third party technology, such as leading algorithms for matching against large databases. In this case, a capture device is required with the flexibility to provide biometric templates to the third-party algorithm for matching, or to run a third-party template generation algorithm on the device. Each of these system architectures has its own advantages.

SMBs to chains and global enterprises

Biometric database storage and matching directly on a device like the Intel RealSense F450 and F455 allows companies to use high-performance 3D face authentication of up to 1,000 people with an on-device database out of the box. This means a small business can set up face authentication to enhance the security of its workplace without further investments in hardware or software. The RealSense ID API runs on the business’ existing host controller.

For businesses that want to store their database on the host controller, for instance to run on several kiosks or for a larger database, the system can be implemented to perform biometric template generation and anti-spoofing on the device, with image matching on the host.

A face authentication system for applications like PoS devices or ATMs will require a server or cloud architecture, with the API delivering the template of the liveness-detected face for matching in the back end.

A hotel implementing face authentication for its customers could offer check-ins through a kiosk in the lobby or at the front desk, running on the host, or through the guests’ smartphone via the cloud. Then access to the hotel room could be controlled with face authentication against the database on the terminal host.

3D face authentication systems can also include third-party biometric matching algorithms running either at the host level or in the cloud. In the former case, the third-party algorithm can use the image from the 3D camera, and compared with a database stored in the cloud and distributed to terminals. This kind of implementation allows the organization to use the algorithm of its choice while keeping control over the matching environment, which may help with privacy or regulatory compliance considerations.

For a third-party algorithm in the cloud, the biometric template will be generated by software running on the host, which is sent to the cloud for matching. Businesses chose this kind of system to make use of larger cloud solutions or networks including facial authentication and requiring a particular algorithm.

Intel RealSense cameras provide the flexibility to support any of the above implementations, allowing businesses to architect their systems based on their particular requirements. With recent advances in edge processing and imaging, whether for convenient, touchless access control in a small business or secure payments and loyalty programs at a national chain, the benefits of 3D biometric facial authentication are now easy to integrate.

Related Posts

Article Topics

 |   |   |   |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Research

Biometrics White Papers

Biometrics Events

Explaining Biometrics