Multi-modal Biometric Feature Identification Research Based On Fingerprint And Palm Print

Nowadays，the biometric feature identification technology has gradually replaced thetraditional identification techniques in our daily life. Compare with traditionalidentification techniques, single-modal biometric feature identification is more secure andaccurate，but it also has the disadvantage of the large affection by sensor noise, low generalapplicability, weak performance to anti-counterfeiting attack. Multi-modal biometricfeature identification is the identification which uses a variety of biometric feature as thematching reference, it can solve a large degree of the exposed problem.This article is the research of multi-modal biometric feature identification based onfingerprint and palm print. First, fingerprint and palm print are preprocessed, and then meltthe images of the two kinds of biometric feature, finally, extract feature vector from thenew synthetic template for matching recognition, This article is different from the ordinarymulti-modal feature identification method, not separate the whole biometric featureidentification system, and take use of the results of their respective systems. It combinestwo biometric features to generate a new feature vector, and use this vector to matchingrecognition.This article is about all the processing of a identification system except imageacquisition. During palm print preprocessing, this paper propose an adaptive interceptionmethod of palm print ROI. During fingerprint preprocessing, this paper proposes a newdirection filtering algorithm different from the traditional algorithm based on8directions.This algorithm fit fingerprint ridge direction more appropriate. What’s more, a newlocation method of center point fingerprint based on simple directional field is proposed.And present a new biometric feature fusion method: composite the images of fingerprintand palm print after preprocessing, extract feature vectors and match recognition based onthe new template. Experiments show that this identification method is feasible andeffective.