Research On Fingerprint Identification Algorithm And Its Hardware Design

As the fingerprint is unique and invariant, as well as the fingerprint recognition technology is feasible and practicaly, fingerprint identification has become the most popular, most convenient and most reliable personal identity authentication technology. As the technology advances, people also put forward higher requirements to the Fingerprint Identification System. Accurate and fast fingerprint identification system has always been a hot issue of research.The algorithm of pre-processing, post-processing and matching determine the performance of the system in a large extent. Thus this dissertation mainly focuses on the algofithm of them.Moreover this dissertation designs a new matching algorithm. The jobs completed in this dissertation are as follows:Firstly, fingerprint image algorithms are studied.After considering the effect of image processing and hardware costs, I choose the gabor filter for image enhancement. Moreover its C program in float point and fixed point is written. And the pre-processing ASIC module is designed.Secondly, fingerprint image post-processing algorithms, including ridge tracking, spur mending, bridge mending, hole mending, short ridge and broken ridge mending, are studied and compared. The algorithm is implemented with C language in both float point and fixed point, and Verilog code.Thirdly, the disertation study the center-based matching algorithm, and designs a new matching algorithm base on it. This algorithm's matching speed is fast, the capcity is small.A serial test is done based on standard fingerprint database and self-making fingerprint database for float point and fixed point C program and Verilog code. Moreover the disertation use verilog establish the pre- and post-processing algorithms's behavioral modeling.