Method, device and computer program for detecting point correspondences in sets of points
A method, device and computer program for detecting point correspondences in sets of points and in particular for fingerprint verification. Points of note on fingerprint lines in a scanned fingerprint are compared with corresponding points of note on fingerprint lines in a reference fingerprint and matching pairs are formed from possible corresponding points of note in the scanned fingerprint and the reference fingerprint. A maximum number of such matching pairs is found. This allows an efficient method to be specified, for fingerprint verification for example, that requires only a small amount of working memory and computing power.
1. A method of detecting point correspondences between a first set of points and a second set of points, comprising the following steps:
finding of possible matching pairs comprising a point from the first set of points and a point from the second set of points, and
finding a maximum number of matching pairs, wherein the finding of a maximum number of matching pairs comprises the following steps:
combining of two matching pairs at a time into a tuple to form a plurality of tuples,
determining an affine map for each tuple, that maps the two points in each matching pair of the tuple onto one another,
sorting of the plurality of tuples into a tuple list,
creating a search tree on the basis of the tuple list, and
deriving a matching result from the search tree;
wherein the point from the first set of points and the point from the second set of points are each points of note on fingerprint lines, the first set of points corresponding to a scanned fingerprint, the second set of points corresponding to a reference fingerprint, the points of note on the fingerprint lines in the scanned fingerprint allowing to be matched with the points of note in the reference fingerprint, and the method being a method of fingerprint verification in which the scanned fingerprint is compared with the reference fingerprint.
2. A method as claimed in claim 1 , wherein the sorting of the plurality of tuples into the tuple list comprises the following steps:
identifying valid tuples among the plurality of tuples by checking the plurality of tuples for validity,
assessing the valid tuples by a cost function to determine costs for the valid tuples,
sorting the valid tuples and sorting the valid tuples into the tuple list, the valid tuples being sorted into the tuple list in ascending order by cost.
3. A method as claimed in claim 1 , further comprising the following steps:
determining a match rate and a hypothesis rate,
forming a two-dimensional result space in which the match rate is entered along a first dimension and the hypothesis rate along the second dimension, and
finding a point in the two-dimensional result space on the basis of the match rate and the hypothesis rate,
forming a cluster in the result space, and
finding a matching result on the basis of a distance of the point from the cluster.
4. A device for detecting point correspondences between a first set of points and a second set of points, comprising:
a processor for performing the steps of finding possible matching pairs comprising a point from the first set of points and a point from the second set of points, and
finding a maximum number of matching pairs, wherein the finding a maximum number of matching pairs comprises:
combining two matching pairs at a time into a tuple to form a plurality of tuples,
determining an affine map for each tuple, that maps the two points in each matching pair of the tuple onto one another,
sorting the plurality of tuples into a tuple list,
creating a search tree on the basis of the tuple list, and
determining a matching result from the search tree;
wherein the point from the first set of points and the point from the second set of points are each points of note on fingerprint lines, the first set of points corresponding to a scanned fingerprint, the second set of points corresponding to a reference fingerprint, the points of note on the fingerprint lines in the scanned fingerprint allowing to be matched with the points of note in the reference fingerprint, and the device being a mobile fingerprint verification device in which the scanned fingerprint is compared with the reference fingerprint.
5. A device as claimed in claim 4 , wherein the sorting the plurality of tuples into a tuple list comprises:
a processor for performing the steps of identifying valid tuples among the plurality of tuples by checking the plurality of tuples for validity,
the valid tuples by a cost function to determine costs for the valid tuples, and
sorting the valid tuples and sorting the valid tuples into the tuple list, the valid tuples being sorted into the tuple list in ascending order by cost.
6. A storage media that stores a computer program for a fingerprint verification device, wherein point correspondences are detected between a first set of points and a second set of points, and wherein computer program code of the computer program, when executed by a processor, performs the following steps:
finding of possible matching pairs comprising a point from the first set of points and a point from the second set of points, and
finding a maximum number of matching pairs, wherein the finding of a maximum number of matching pairs comprises the following steps:
combining of two matching pairs at a time into a tuple to form a plurality of tuples,
determining an affine map for each tuple, that maps the two points in each matching pair of the tuple onto one another,
sorting of the plurality of tuples into a tuple list,
creating a search tree on the basis of the tuple list, and
deriving a matching result from the search tree;
wherein the point from the first set of points and the point from the second set of points are each points of note on fingerprint lines, the first set of points corresponding to a scanned fingerprint, the second set of points corresponding to a reference fingerprint, the points of note on the fingerprint lines in the scanned fingerprint allowing to be matched with the points of note in the reference fingerprint, and the method being a method of fingerprint verification in which the scanned fingerprint is compared with the reference fingerprint.
7. A method of detecting point correspondences between a first set of points and a second set of points, comprising the following steps:
finding of possible matching pairs comprising a point from the first set of points and a point from the second set of points, and
finding a maximum number of matching pairs, wherein the finding of a maximum number of matching pairs comprises the following steps:
combining of two matching pairs at a time into a tuple to form a plurality of tuples,
determining an affine map for each tuple, that maps the two points in each matching pair of the tuple onto one another,
sorting of the plurality of tuples into a tuple list,
creating a search tree on the basis of the tuple list, and
deriving a matching result from the search tree,
determining that the first set of points corresponds to the second set of points if the matching result is above a presettable threshold value,
using the matching result for biometric identification, and
outputting a result of the biometric identification to a display.