Accelerating Fingerprint Identification using FPGA for Large Scale Applications

Abstract

Real time fingerprint identification for large scale applications is a very challenging task not only from computations point of view but also in terms of accuracy.In this research both these challenges are addressed and accurate minutiae matching algorithm specially designed for hardware based accelerator is proposed. The research has several contributions; An accurate fingerprint matching algorithm based on minutiae pattern matching is initially implemented and tested on CPU. Modifications are made in the algorithm in order to make sure efficient use of hardware resources without any significant effect on accuracy. The RTL design for the algorithm is developed in order to port it to an FPGA. Further study is done to optimize the RTL level hardware design so that it consumes fewer resources and works efficiently for real-time implementation in larger AFIS. As a result, a hardware accelerator for minutiae pattern matching, that is capable of performing multimillion matches per second is developed. The design is scalable and can be customized according to the application needs. To the best of our knowledge, the design outperforms other fingerprint matching design implemented on FPGA. It is ensured through rigorous testing that the fingerprint-matching unit is not only fast, but it also gives accurate matching results. In future, as an extension of this work, a complete system is envisioned to be developed that involves real-time fingerprint acquisition, enhancement, feature extraction, encoding and matching with database templates.