| The fingerprint identification is one of the most reliable, convenient and popularpersonal verification technologies due to its uniqueness and invariability. Based on theresearch into extensive literature of automatic fingerprint identification system, this thesisfocuses on the algorithms of fingerprint identification system and the implementation ofsuch techonology within file encryption system. The results show that the proposedalgorithm is valid and feasible. The main contributions of this thesis are summarized asfollows:The algorithm of automatic fingerprint identification is studied. This thesis conductedthrough analysis and comparison with the individual steps of the pretreatment process,including fingerprint segmentation, enhancement, binarization and thinning, and furtherproposed a new image segmentation algorithm. The feature extraction of fingerprint isstudied. This thesis adopts a8-neighborhood algorithm to extract minutiae, so thatminutiae can be extracted with increased accuracy. The paper utilizes the spuriousminutiae’s structural features to remove them. The fingerprint feature matching section isinvestigated. This thesis proposed a fingerprint-matching algorithm based on the vectortriangle, which can determine the reference points in fingerprint image and then realize thematching by utilizing information of two nearest minutiae.The algorithms for automatic fingerprint identification have been realized,debugged and tested successfully with Visual C++6.0Builder on PC. All algorithmsmentioned have been tested in FVC2004standard fingerprint database, and theexperimental results prove a significant increase of the recognition rate.The application of fingerprint identification technology is important for fileencryption system, using fingerprint key to encrypt and decrypt, which will help toimprove data safety. The results show that the proposed algorithms for fingerprintindentification are valid and feasible. |
PDF Full Text Request