A Novel Fingerprint Matching Method Combining Geometric and Texture Features

In this paper, we proposed a new fingerprint matching algorithm based on local geometric feature of fingerprint minutia and texture feature for each minutia. To describe the geometric feature of fingerprint minutia, we build a bi-minutia based bar model and get the geometric relationship between the bar and ridges of two candidate minutia; to demonstrate the texture feature, we creatively adopt gradient angular histogram in the neighborhood region of minutia. Meanwhile, changeable sized boundary box of unique area adopted for minutia matching make this algorithm more robust to nonlinear fingerprint deformation. Finally, experimental results on the database FVC2004 demonstrate that our method is effective and reliable, whilst the matching accuracy can be improved to some extent after using gradient angular histogram as texture feature without adding extra amount of calculation.