A Deep Dive into TENSOR’s Technological Innovations: Fingerprint Biometrics
The TENSOR project proposes a new method of addressing current challenges regarding the safe, quick, and robust exchange of biometric data in cross-border criminal scenarios. To achieve this goal, an important cornerstone of the project is the implementation of a variety of different biometric modalities.
While Fingerprinting is a modality of biometric identification with a long history and proven track record, there are difficulties, especially considering an environment with a need for quick answers in complex settings, often across different countries.
Common methods of stationary Fingerprinting require a process of either the change of the physical location a check was initiated in, meaning a suspect has to be transported to a Police station for instance. Other methods include specific hardware that is bulky and expensive, which mobile teams need to acquire and carry with them.
Therefore, in TENSOR, we will provide a mobile Fingerprinting solution that captures fingerprints with the smartphone camera, utilizing the devices Police forces already carry, without the need for additional hardware. The fingerprints taken with the smartphone camera adhere to NIST and other international standards.
While recording with the smartphone camera, images are taken from the recording through a best shot analysis and those images are converted into a 500dpi output format, which is compatible with international AFIS databases of flatbed enrolled fingerprints. Using a backend service for matching and forwarding to storage, integrated into the TENSOR platform, those fingerprints are taken in-field, and because they adhere to international standards, they can immediately be matched against AFIS databases. Instant matching results are displayed on the officer’s device, as well as stored on the TENSOR platform for later reference.
To make storing of biometric data on the TENSOR platform possible, the consortium provides data protection and privacy components, such as homomorphic encryption of data, and it strives to develop multi-modal deep hashing architectures to ease large-scale biometric data indexing. The TENSOR platform will also improve efficiency through interoperability with legacy systems, while providing novel secure means of biometric data exchange capabilities.
Enabling users of the TENSOR platform to combine different biometric modalities in their investigations, including a Fingerprinting solution, which has a very low hurdle to use and produces instant reliable results, allows for more streamlined cross-border investigations. Once they have their suspect in front of them, with the TENSOR Fingerprinting solution, an officer simply needs to use their smartphone to quickly collect the suspect’s fingerprints, match them against a national or international database on the spot, and save them to the TENSOR platform at the same, allowing them a quick way to also address international agencies, without delay, and in a secure manner.
Next to providing privacy and security, the TENSOR consortium works together to define practical workflows including all biometric modalities, such as face, gait and voice biometrics, and to fuse the results of each individual check into gaining a broad picture of one or several suspects, providing investigators with invaluable information that is centralized and easy to access.
In the TENSOR project, we aim to create a best practice scenario for future-proof biometric identification and re-identification. In the case of Fingerprinting, the use-case scenarios will provide insight into how law enforcement will benefit from a mobile method of Fingerprinting – from an economic perspective due to the disregard of extra hardware, from a timing perspective due to the disregard of transportation of persons, from a data security perspective due to the advanced methods of the TENSOR platform, but also from a very practical hygienic perspective, by providing a touchless method.
