Fingerprint Automation Identification System Based On FPGA

Because of the advantages of uniqueness, stability, easy availability in fingerprint, the finger identification technology has become the most widely applied techniques in biometric verification technology. However, the traditional Automatic Fingerprint Identification System (AFIS), more based on PC, doesn't meet the requirements of portability and real-time performances in its use, because of big cubage and complicated algorithm, and just this affected its application development and popularization. Moreover, the problem of False Match and False Non-Match is common in the present AFIS. Today, the pre-processing of fingerprint image and fingerprint match is the ever-challenging job in image process and pattern recognition.This dissertation describes the realization of AFIS with Filed Programmed Gate Array (FPGA). Nowadays, the density of FPGA on a chip is about one million logic elements. With the prize decreasing, it is feasible and afforded to design AFIS based on FPGA. Under the concepts of the Altera SOPC, this system uses Nios II embedded processor and FPGA hardware module to achieve most algorithms of AFIS. During pre-treating fingerprint image, it needs processors to process the large flow of data quickly. As the architects of the processors are simple usually, this dissertation uses FPGA hardware modules to achieve them. On the contrary, this dissertation uses Soft-Core (Nios II) to achieve the complicated algorithms, which use to process the lower flow of data when system processes the information after pre-treatment such as fingerprint match..Under consideration of performance, to minimize resource need to improve the present pre-treatment algorithms of fingerprint image before they are implemented by FPGA hardware modules. Refer to acquirement of ridge orientation of fingerprint, a segment of the pre-treatment of fingerprint images, we use CORDIC to implement trigonometric function. Refer to complication of finger enhancement based on Gabor filter, it is decomposed into 1-D Gaussian filters along orthogonal axes firstly, one is implemented with template then, and the other one is implemented through rotation of the template. Refer to the other segments of the pre-treatment of fingerprint images, they are all achieved using FPGA hardware modules with referring to the have-got algorithms. And experimental results indicate that the processing speed is increased with the same reliability using FPGA hardware modules. Refer to algorithm of the fingerprint match, a new criterion and the concept of variable limits box are posed on the traditional algorithm which is based on point pattern to implement the fingerprint match.Experimental results show that the proposed framework of the system is effective and feasible. AFIS based on FPGA completes the functions of the traditional, but it has unique advantages in terms of speed and scalability. The completion of this dissertation grounds for the in-depth research of AFIS.