Research On Automated Fingerprint Identification System

Due to their uniqueness and persistence, fingerprints are used as main basis of personal identity. Automated fingerprint identification system, a technology of fingerprint identification using computer, is of convenience, high efficiency,security and reliability. It has been applied in many fields such as financial security, data encryption, electronical business and will play a more and more important role in our life.Based on latest extensive discourses and technology journals in this field, the dissertation is trying to make studies on algorithms of AFIS . Mainly works are as following:(1) Fingerprint preproces. Its purpose is removing noise of fingerprint image and improving definition of fingerprint image ,get the structure of fingerprint ridge.change fingerprint image to a one pels width image .Which will be beneficial to the follow feature extraction . Preproce includes image normalizing, orientation estdmation,frequence estimation ,ridge extraction ,ridge removed noise, thinning .(2) Fingerprint feature extraction. Its purpose is get all true ending and bifurcation. The two kind of minutiae can be extracted by using 8-neighborhood cross-calculation algorithm. A feature extraction postprocessing algorithm would delete all false minutiae.(3) Fingerprint image matching. Its purpose is judge whether the two fingerprint are captured from the same finger based on their minutiae. We advanced an algorithm . Screen the primary reference points and remove those wrong points according to the property that two identical minutiaes will necessarily creat a similar trigon with two points nearest to each respectively. Two identical minutiae lead to identical ridge number to their respective reference point, which is considered as one of the conditions to determine whether or not the two reference points are identical. The experiements showed that the can effectively avoid the error from the rotation and nonlinearity of the fingerprint.The algorithm is tested in FVC2000 DBl,the true accept rate is 91.03%. the false reject rate is 7.91%. the false accept rate is 1.06%, the time of comparing two fingerprint is 0.019s.