The invention discloses a deformed
fingerprint matching method based on a curvilinear coordinate
system and transference of reference
minutiae. The method comprises the following steps: (1). a
fingerprint image is represented that I(m*n),O(i, j) belonging to [0, pi) is the local
ridge orientation at the pixel (i,j), and each minutia in the
fingerprint image I is described by coordinate, direction and type; (2). X axis and Y axis of the curvilinear coordinate axes are extracted from each minutia by taking each minutia as an origin based on a
ridge mode and an
orientation field of the
minutiae; (3). scales of the curvilinear coordinate axes are computed; (4). the coordinate relation between every two
minutiae mi and mj are computed; and (5). The minutiae are matched by the method based on the transference of the reference minutiae: provided that a template
fingerprint image T and an input
fingerprint image Q contain p minutiae and q minutiae respectively, C(T) equals (Ci, j<T>), C(Q) equals (Ci, j<T>), wherein, i is not less than 0 and j is less than q; T and Q are matched by utilizing the minutiae of the C(T) and C(Q) pair to compute the matching minutiae. The method has the advantages of high identification accuracy rate, high efficiency and
simple algorithm.