The Hardware Acceleration Technology Research Of The Algorithm Of The Application Of Fingerprint Image Recognition

Fingerprint recognition system is widely used in the daily life and the fingerprint feature extraction and fingerprint feature matching are two key processes in the system.The Hardware Acceleration Technology Research of the key algorithms of the orientation field model base on Sparse Coding and the fingerprint matching algorithm based on minutiae is studied.Firstly,the key of the orientation field model base on Sparse Coding is considered.The hardware structure of the Sparse Coding algorithm is proposed and the fixed-point Sparse Coding algorithm is evaluated.Then the paper implemented the hardware structure.Experimental results show that the structure can accelerate the operation of the Sparse Coding algorithm and maintain the accuracy of the algorithm.Secondly,according to the fingerprint matching algorithm based on minutiae,the paper proposes a two level fingerprint matching accelerator.And the first level fingerprint matching accelerator is a fully pipelined architecture and adopts fine-grained pipeline to ensure the parallelism of the computation.The second stage accelerator is based on the first stage and adopt a two-level-parallel structure and multi-channel storage scheduling to enhance the throughput of the fingerprint matching algorithm.The experiment results show that the fully pipelined accelerator can acquire the throughput of 430000 pairs of fingerprints per second.And the coarse-grained pipelined accelerator can achieve 9.63 millions per second.Finally,based on the proposed structures,this paper presents a FPGA parallel array structure,which is applied to fingerprint recognition.The communication structure is optimized,and the feature extraction algorithm and feature matching algorithm are tested on the FPGA array.The experimental results show that the fingerprint image processing system based on the FPGA array can adapt to the real time requirement of the application of the hundreds of millions of magnitude fingerprint processing.