Fingerprint Matching Using Texture Feature Extracted from Minutiae Neighborhood

Abstract

The paper proposes a method of extracting a rotation-translation-invariant texture feature from minutiae neighborhood to help matching fingerprint minutiae. The texture feature is extracted by complex filter, and represents pattern of orientation image around minutiae, and is used as one element feature of minutiae. Based on the texture feature, a new and simple texture matching distance differed from the average Euclidean distance is defined. This texture feature is easy integrated into existing minutiae matching algorithm. Experimental test results show that the texture feature is able to distinguish minutiae with different orientation images in neighborhood, and the texture matching distance defined has better discrimination than the average Euclidean distance, and the texture feature has an important role to accelerate the matching speed and improve the matching accuracy when it is integrated into fingerprint matching.