• Finger-print based student attendance register

      Parvinzamir, Farzad; University of Bedfordshire (University of Bedfordshire, 2012-09)
      Monitoring student attendance in the UK has become a prime concern for Universities in recent months, due to a perceived lack of accuracy in reports submitted to the UK Borders Agency and political pressure about wider immigration issues. This project proposes a biometrics-based solution to that concern which also conforms to legislative pressures on data governance and information security, but which can provide accurate, reliable data for the institution to use in future reports to UKBA. All biometric techniques obviate the need to carry a token or card, or to remember several passwords, and reduce the risk of lost, forgotten or copied passwords, stolen tokens or over the shoulder attacks. This project shall focus on using fingerprint recognition, mainly due to the low-cost of devices for deployment and high user acceptance. Fingerprint recognition has traditionally been used for data access amongst a mobile population with increasingly portable devices, but it can also be employed for monitoring purposes, and this project defines how it could be used in this context to provide a fingerprint-based student attendance register. This project set out to overcome the drawbacks of the current attendance system, which can be fooled by “buddy swiping” of absent students’ RFID card or signing the register sheet on behalf of absentee students within a university. An application was designed within MATLAB to identify pattern in data, extract vectors from a fingerprint image and map values to the new area, then to verify a student who swipes his fingerprint against those values. The requirement was to make this system work asynchronously so that constant internet and database connections are not required, to deliver outstanding rates of accuracy, and to ensure this could work on machines with very low computing power so that it can be utilized in mobile devices in future. The delivered application uses the Principal Component Analysis method to compare fingerprints with the new form of harmonized data defined by eigenvectors and eigenvalues in n dimensions. This high-speed method uses the lowest computational power to deliver accurate results through making a closest match against stored values. This application has potential to be employed as a modular add-on by a University student monitoring system or connect to its database and transfer data.