The Design Of Fingerprint Image Acquisition System Based On DSP

With the development of information technology, the security and reliabilityof adopted ID-recognition technology has been the most crucial factor in systemdesigning. As traditional ID-recognition methods, such as password and IC card,can't meet the growing demands, biological recognition technology becomes themost convenient and the safest method for the reason that the character ofbiology is unique and invariable in lifetime. With its universality, permanence,attainable data acquisition, easy implementation and acceptability, fingerprintrecognition technology has become one of most effective biological recognitiontechnologies. In addition, with lower cost of system comparing with otherbiological recognition technologies, fingerprint recognition technology has beenwidely used in applications, such as gatekeeper system, attendance checkingterminal, insurance system, electronic commerce and so on.Firstly, the design and implementation of fingerprint data acquisition systemis discussed in this thesis. The characteristic, configuration, basic function andprocessing principle of the fingerprint data capturing chip named MBF200 areintroduced, as well as some main characteristic of the DSP evaluation board. Theinterior configuration, processing principle and working method of the chip areanalyzed. Then based on the knowledge of the communication method of EMIF,which is the interface between DSP and extern memory, a scheme of fingerprintdata acquisition based on MCU interface mode was proposed. Also some moredetails are given in this thesis such as linking method of hardware system, PCBof data acquisition sub-board, software flow chart of fingerprint data acquisitionand debugging process.Secondly, fingerprint image preprocessing, feature extraction and matchingmethods based on the detailed characteristic are studied and implemented in thisthesis. Including Pre-filter, extreme-value filter, directional diagram obtained,binary conversion, slendering, detailed point features extraction, removal of falsecharacteristic points, the central point extraction and the matching etc. Thisthesisdetailed the more important parts include: the method to obtain point directionand block direction in the preprocessing; Feature extraction and the process to removal of false characteristic points in matching and the algorithm of centerpoint extraction etc. Because the handling effects of these parts affect fingerprintmatching results directly.Finally, identified the algorithms selected above on the platform of Matlaband it includethree parts.The first one is that two preprocessing algorithms flowand the corresponding results were lay out. The second one is that throughchecking and analysis to processing effect of different quality fingerprint,fingerprint image preprocessing scheme fir for the data acquisition system in thisthesis was selected. And the last one is the confirmation results of the matchingalgorithm based on the detailed point has been lay out and analyzed, andcorresponding improvement method was proposed.