|
The uniqueness of fingerprints, universality and invariance，established the importantposition of fingerprint in identity authentication, fingerprint recognition technology has beenwidely used in various aspects of social life. With advances in technology, fingerprintrecognition system is miniaturization, low power consumption, high performance, andembedded development.Based on the recent years on the fingerprint recognition technology, combined with DSPtechnology, the existing fingerprint recognition algorithm has been improved, and build aDSP-based embedded fingerprint identification system.In hardware design,this paper shows ahardware block diagram of the whole fingerprint recognition system, and for each one of themodules are introduced,the core of this system uses the TI company TMS320VC5501DSPchip with high performance,which has strong data processing ability, and can meet thereal-time requirement of the system.Fingerprint acquisition module optional fingerprint sensorTI’s TFS-M12(B), completed the acquisition of the fingerprint image.In algorithm design,based on the hardware platform, the development of embedded fingerprint recognitionalgorithm, including: fingerprint image preprocessing, feature extraction and matching.Thispaper makes a deep analysis and Research on the process of fingerprint imagepreprocessing,including normalization, image segmentation, binarization, noise removal,thinning algorithm.Image binarization algorithm uses a binarization method based onorientation field, use the directional characteristics of the fingerprint image, gray change, todetermine the dynamic threshold of the fingerprint image, the fingerprint imagebinarization.In this paper, the characteristics of the fingerprint image matching algorithm hasbeen improved, divided into two phases, the first phase according to the extracted featurepoints, construction of similar triangle, by retrieving similar triangles to determine thematching score, the second stage is the matching algorithm based on local feature points,according to the number of feature points matching to determine the matching result, throughthe stages of screening out part of fingerprint matching, reduces the calculation quantity.Finally, the entire fingerprint identification system testing, and analysis of theexperimental data, False Accept Rate and False Reject Rate of the whole system are withinthe acceptable range, the matching result has achieved the anticipated effect.
|