Research On The System Of Fingerprint Identity Based On DSP

Fingerprint has been fully used in fields which contain identity authentication with its universality,uniqueness and invariability.Fingerprint identification technology based on biological characteristics,which relates many technologies,including pattern recognition, digital image processing,artificial intelligence,computer and so on,has not only theoretical significance but also practical value.Most of the traditional fingerprint identification systems are based on PC platform.With the development of technology and the demand of market application,people demand more for fingerprint identification system. Fingerprint identification system is developing towards miniaturization and embedded.In this paper,hardware design and software design are described briefly,fingerprint identification algorithms are analyzed and studied in detail.About hardware design aspact,the whole system hardware frame is given firstly,erery part is described in detailed.The core of the system is DSP chip made in TI company.Using its powerful ability on data processing,system can quickly complete kinds of tasks assigned by proeessing algorithm for fingerprint image and meet to the request of realtime charaeteristic.About software aspact,the initializations of DSP and fingerprint sensor MBF310 are introduced,fingerprint images are collected.About algorithm aspact,through comparing all algorithms,this paper selects fingerprint recognition algorithms relatively suitbale for DSP hdarware system,with combining their own characteristic of the fingerprint imgaes.Algorithms about fingerprint image pretreatment,minutiae extraction and fingerprint matching are introduced.The corresponding algorithms about fingerprint matching are improved,a subsection matching algorithm based on structure minutiae is adopted.This algorithm constructs feature vector according to the feature that the amount of bridges between minutiaes is unchanged.After the first stage of subsection matching,the problem of selecting the reference point is solved.The computation of the second stage is reduced by removing some fingerprints.At last,the debugging of the system and simulation of the algorithms are achieved and the experimental data is analysed.The FRR and FAR of the whole system are acceptable.The subject research has certain reference value for the product development and application of embed fingerprint identification system.