Innovatrics is proud to announce exceptional level of performance and generally excellent results of its AFIS algorithm in the Fingerprint Vendor Technology Evaluation (FpVTE) 2012.
The FpVTE, organized by the National Institute of Standards and Technologies (NIST), is the most important evaluation of large 1-to-many fingerprint identification algorithms using enrollment sample sizes in the multi-millions. It is the first independent evaluation focused on large scale fingerprint identification since FpVTE 2003.
The FpVTE report classifies Innovatrics’ submission among the best in every evaluation scenario when ranking overall performance.
The overall performance of submissions is expressed as an operational score combining the most relevant identification statistics. Factors incorporated into operational ranking were: Identification Time, Accuracy, Enrollment Time and RAM Consumption.
These criteria depict the most relevant AFIS (Automated Fingerprint Identification System) requirements.
The Innovatrics algorithm is ranked in the top performance group (result in 80th percentile or better) in each of the nine tested scenarios (one-to-many identification using various finger combinations from single finger up to ten fingers).
Besides Innovatrics, only one other vendor consistently scored in the top performance group (see chart below Operational Rank Summary).
The balanced overall performance of Innovatrics algorithms in all categories and across all scenarios predetermines Innovatrics as a universal provider of complete solutions for fingerprint identification.
This proves that our product portfolio integrating benchmarked algorithms is suitable for a wide range of applications: real-time applications with one and two fingerprints, civil AFIS solutions for large scale deployment, as well as law enforcement AFIS applications.
The FpVTE evaluation demonstrates that the Innovatrics algorithm achieved high levels of accuracy while processing fingerprints faster than other vendors.
Being fast and accurate are very important factors for large scale AFIS systems as well as for real-time identification solutions. Innovatrics‘ ability to be one of the fastest among the most accurate vendors on the market makes its technology suitable for nation-wide projects with large quantities of records (AFISes for ID cards, passports, etc.), for systems with very high throughput (border control solutions), and real-time identification systems (banking CRM, watch list check, etc.).
The high speed of Innovatrics technology saves hardware resources, reduces maintenance, and makes our product more energy efficient and thus environmentally-friendly.
The charts below summarize Innovatrics’ performance in the most common scenarios related to civil AFIS: Left and Right Index Search and Ten-finger Identification Flats Search.
According to the FpVTE report, Innovatrics has the shortest average enrollment time among all the vendors. Innovatrics enrollment is 70% faster than the second fastest vendor. Enrollment time comprises fingerprint template extraction time from raw image and image segmentation time in the case of multi-finger images. As enrollment is a mandatory step in every AFIS transaction (registration, verification, and identification), fast enrollment is one of the key factors in achieving low latency and high throughput of AFIS operations.
Innovatrics leading position in enrollment time is a result of our long-term experience as a provider of SDK modules for fingerprint scanner manufacturers. SDK modules are often used on a wide range of platforms (embedded modules, mobile solutions, smartphones, etc.) where extraction speed is crucial.
The Fingerprint Vendor Technology Evaluation (FpVTE 2012) – a worldwide comprehensive range of tests that is open to everyone in the industry – was conducted at the National Institute of Standards and Technology (NIST). These tests were primarily performed to assess the current capabilities of fingerprint matching algorithms using operational data-sets containing several million subjects.
The tests included three classes of participation that examined one-to-many identification using various finger combinations from single finger up to ten fingers.
Enrollment sets used for one-to-many identification varied in size from 5,000 up to 5,000,000 enrolled subjects. Any segmentation of four-finger slap images or two-thumb captures was performed by the submitted software. All data used was sequestered operational data that was not shared with any of the participants.Read more