The MegaMatcher 11 line of biometric on-premise and cloud solutions and SDKs includes significant multimodal biometric algorithm enhancements and NIST-proven accuracy, speed and interoperability.
Neurotechnology, a provider of deep learning-based solutions and high-precision biometric identification technologies, today announced the release of the new MegaMatcher 11 line of biometric solutions, cloud technologies and software development kits (SDKs). The latest versions include major algorithm enhancements that improve accuracy, speed and interoperability across multiple biometric modalities: fingerprint, face, iris, voice and palmprint – any of which can be used alone or in combination. The algorithms in the new MegaMatcher 11 product line – which are among the top ranked in the NIST MINEX, PFT, FRVT and IREX evaluations – are available in a variety of formats to meet any biometric identification needs. For projects that require a turnkey biometric solution, Neurotechnology has updated their complete MegaMatcher Automated Biometric Identification System (ABIS) – available as either on-premise software or a new cloud service – with all of the necessary components for deploying a fully operational, large-scale biometric system. For developers and system integrators, Neurotechnology offers MegaMatcher SDKs and Accelerators that enable the incorporation of their fast and reliable algorithms into new and existing biometric systems for either small or large-scale uses.
“With continued research in modern neural networks, our team introduces another leap in person identification accuracy by using multiple biometric traits separately or together,” said Dr. Justas Kranauskas, R&D Manager for Neurotechnology.
NEW ALGORITHMS PROVIDE GREATER SPEED, ACCURACY AND INTEROPERABILITY
The algorithm enhancements in the MegaMatcher 11 product line include fingerprint, face, iris, voice and palmprint modalities:
Fingerprint algorithm updates for template generation, extraction, classification and matching improve accuracy, speed and interoperability. The classification algorithm improves both classification accuracy and speed and a new slap image segmentation algorithm detects the orientation of each fingerprint, allowing for the handling of more complex cases.
Neurotechnology’s fingerprint algorithms were top-ranked in the latest NIST evaluations:
- NIST MINEX III: Neurotechnology’s template generators are ranked as the most interoperable MINEX III compliant template generators on the market. The MegaMatcher 11 template generator and matcher pair are ranked the 2nd most accurate native MINEX III compliant pair.
- NIST Proprietary Fingerprint Template Evaluation II (PFT II): Neurotechnology’s one-to-one fingerprint verification algorithm (submission 4E) is the 2nd most accurate algorithm on average across all 33 experiments.
Face detection and recognition algorithms are significantly enhanced in MegaMatcher 11. The latest face detection algorithm locates many more faces in varied and difficult conditions and can detect up to a full profile rotation of the head. Compared to the previous version, face recognition accuracy is five times higher when set at a low False Acceptance Rate. Other face processing algorithms were also updated, including: more accurate facial feature points detection, better face image quality estimation and improved face properties and emotion classifiers.
The latest facial recognition algorithms also achieved high scores in NIST FRVT evaluations:
- FRVT 2018: Neurotechnology’s face identification algorithm is ranked among the six most accurate algorithms (out of 40 different vendors) in the largest test on a population of 12 million subjects and is ranked 3rd (out of 40 different vendors) when face photos were taken up to 18 years after the initial enrollment photo.
- FRVT Ongoing: Neurotechnology’s face verification algorithm is ranked among the 10 most accurate algorithms (out of 56 different vendors) in the mugshot test.
Iris technology updates in MegaMatcher 11 include a new iris detection confidence feature to quickly verify the presence of an iris in a captured image. To help prevent attempts at spoofing through the use of fake iris textures, the latest version also detects the presence of contact lenses, a possible distortion agent in image capture.
The iris algorithm was also among the leaders in the latest NIST evaluation:
- NIST IREX IX: It is the 2nd most accurate iris recognition algorithm (both in verification and identification scenarios) with an ability to accelerate iris matching up to 50 times faster than the closest competitor without significant degradation in recognition accuracy.
Voice recognition in the latest version incorporates a new, text-independent speaker recognition algorithm, which improves the error rate by nearly two times compared to the previous version.
Palmprint in the latest MegaMatcher line comes with a new extraction algorithm and introduces a faster palmprint matcher.
As noted above, any of these modalities may be used individually or in any multi-biometric combination to provide even greater identification accuracy.
The MegaMatcher 11 product line now also includes simplified, updated support for the latest biometric industry ANSI and ISO standards in both the binary and XML encodings.
MULTIPLE PRODUCT FORMATS MEET A RANGE OF IDENTIFICATION NEEDS AND USER EXPERTISE
Updated products in the MegaMatcher 11 line include:
MegaMatcher ABIS – A complete biometric solution which includes all of the algorithms and software necessary for the deployment of large-scale multi-biometric projects using fingerprint, face and/or iris biometrics. The turnkey biometric system is ready to use, or it can be customized to address particular business needs. Capabilities include identity management, deduplication, adjudication and a latent fingerprint editor among other features. The solution is designed for national ID, voter registration with record deduplication, passport issuance, border control, forensic/criminal investigations, government e-services, social services, banking systems, healthcare and other civil and private applications. The latest MegaMatcher ABIS now comes as both an on-premise solution and a new set of cloud services:
- MegaMatcher ABIS On-Premise Solution – For projects that prefer to perform all of their biometric functions on-site, MegaMatcher ABIS is available as ready-to-use software that runs on regular hardware and can be used in conjunction with cloud features as needed.
- MegaMatcher ABIS Cloud Service – With this new release, the MegaMatcher ABIS complete system is now also available as a secure cloud service. Integrated cloud service functionality makes the system more accessible from multiple platforms and locations, enhancing its use as a solution for large and national-scale projects.
Aside from the major algorithm improvements in the MegaMatcher 11 line, the latest MegaMatcher ABIS introduces workflow enhancements for most common usage scenarios as well as a dynamic post-processing of matching results utilizing given biographic data. This feature allows filtering of results based on attributes like location, gender, age, nationality and others.
“Biometric systems tend to be associated with a high level of complexity,” said Denis Kacan, MegaMatcher ABIS product manager. “With the MegaMatcher ABIS product we aim to minimize the possible challenges by delivering a solution which is easy to comprehend, use and maintain. The latest version brings a new level of efficiency with a simplified and adjustable system workflow. The new MegaMatcher ABIS Cloud Service is one more step towards our goal – it can shorten project delivery time and reduce hardware and maintenance costs.”
Among other major projects in 2018, MegaMatcher ABIS was used for deduplication of more than 46.5 million multibiometric voter records in less than two months to prepare for the Democratic Republic of the Congo elections; more than 5 million duplicates were identified.
MegaMatcher Accelerator – MegaMatcher Accelerator is a combined software and hardware solution that provides high-speed, high-volume biometric identification for national-scale projects. It includes fingerprint, face and iris modalities with an even faster and more robust set of algorithms and failover architecture that utilizes a peer-to-peer network structure. Nodes may also be added dynamically. It is available in several editions, including the Extreme edition, which is capable of matching 1.2 billion fingerprints per second, 1.2 billion faces per second or 700 million irises per second on a single server with a GPU, making it well-suited for large scale biometric projects and short-timeline deduplication processes.
MegaMatcher SDK – MegaMatcher SDK is for the development of large-scale single- or multi-biometric systems for Microsoft Windows, Linux, MacOS X, iOS, Android and ARM Linux platforms. It includes the following biometric modalities as well as SDKs for them: fingerprint (VeriFinger), face (VeriLook), eye iris (VeriEye) and voice (VeriSpeak). The MegaMatcher SDK also includes a palm print modality. Each modality can be used alone or in any combination with other modalities.
MegaMatcher On Card SDK – This matching-on-card technology takes advantage of the latest algorithm updates in the MegaMatcher 11 line. It stores a person’s fingerprint, iris and face templates on a smart card and performs template matching in a microprocessor embedded in the card, instead of matching biometric information on a PC processor.
MegaMatcher 11 products and the entire Neurotechnology biometric product line can be purchased from Neurotechnology and through distributors worldwide. A free 30-day trial is available and, as with all Neurotechnology products, the latest version is a free upgrade for existing customers.