Facial Recognition System
The state-owned enterprise State Information Resources Center “Registru” (ÎS „CRIS „Registru”) is a company subordinated to the Ministry of Information Technology and Communications, which carries out state information resources creation and integration projects. Due to its exclusive rights to provide certain information services and products of national importance, ÎS „CRIS „Registru” is one of the main suppliers of information products and services on the Republic of Moldova’s information technology market.
ÎS „CRIS „Registru” holds the main state information resources, such as the State Register of Population, the State Register of Legal Entities, the National Geographic Information System, the State Register of Transport and the State Register of Drivers
In 2016, ÎS „CRIS „Registru” has started the acquisition of a Facial Recognition System in order to achieve the objectives undertaken within the Visa Liberalization Action Plans. The main objective was to contribute to an improved mobility of the Republic of Moldova’s population while preserving public order and security.
Two components were implemented within this project: the Facial Recognition System (FRS) within ÎS „CRIS „Registru” and the Border Inspection System using the facial recognition component (BIS).
The solution was built on the basis of the COGNITEC technology, which is one of the top performing facial recognition applications, as certified by NIST (National Institute of Standards and Technology).
The Face Recognition System includes the following features:
- Introducing new pictures into the database (image-by-image or by multiple-processing)
- Searching for duplicates in the database by facial analysis
- Identifying a picture by comparing it to 1:1 or 1:n images stored in the database
- To each case can be assigned one or more images—the system is better trained if multiple images of the same person are uploaded
- Data about facial image, eye position, creation date, image source/origin, and picture quality are stored. These data are indicators and vectors that are registered for the person’s biometric signature;
The system has a clustered client-server architecture that allows scaling, high performance and availability even when manipulating databases with an impressive number of images/pictures (up to 20 million). The system was installed on a Hewlett Packard Enterprise hardware infrastructure.
The Border Inspection System includes the following features:
- Capture a picture of the individual in accordance with international standards
- Extract biometric data (pictures) from biometric travel documents (eMRTD)
- Extract data from the machine-readable area of the travel document
- Scan and store the graphic image of the personal data sheet from the travel document
- Compare locally two images (1:1 verification): a picture of the person captured using the system and the image extracted from his/her biometric travel document
- Interconnection with the FRS system within the ÎS „CRIS „Registru” for comparing (1:1 verification) two images and comparing (1:N verification) one image
- Ensure a mechanism for updating control filters (blacklist)
The Border Inspection System was implemented using Microsoft .NET Framework technology in conjunction with Windows Communications Foundation for a service-oriented architecture. Also, Microsoft Internet Information Services was used for the application server and Microsoft SQL Server—for the database.
The Border Inspection System is integrated with the Facial Recognition System via standard interfaces provided by Cognitec. By developing additional interfaces, the interaction with the existing information systems within the Border Police was ensured.
Following the successful implementation of the project, the Republic of Moldova became the first country in the region to use a facial recognition system at the level of State Register of Population. Since 2017 ÎS „CRIS „Registru” identified 100 people with double identity and holding fake passports.
The system is employed in 5 state border crossing points and allows detection of fake passports.
The Facial Recognition System is to be expanded to up to 25 locations.