Processing a fingerprint for fingerprint matching

• US 9,514,351 B2
• Filed: 02/12/2014
• Issued: 12/06/2016
• Est. Priority Date: 02/12/2014
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Boshra, Michael
• Primary Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US14/179,537
• Publication Number:
  US 20150278574A1
• Time in Patent Office: 1,028 Days
• US Class Current: 1/1
• CPC Class Codes: G06V 40/1359 Extracting features related to ridge properties; Determining the fingerprint type, e.g. whorl or loop

Abstract

Processing a fingerprint can include determining one or more optimal weights based on ridge flow angles or ridge flow angle differences. Determination of the optimal weight(s) can be based on predicting a ridge flow angle for each cell in a ridge flow map using one or more neighboring cells. The optimal weights may be estimated so as to minimize error between the predicted and actual ridge flow angles. Alternatively, the optimal weight(s) may be determined using a predicted ridge flow angle difference for each cell in a difference map that is based on an actual ridge flow angle difference for one or more neighboring cells. The optimal weights can be estimated to minimize the error between predicted and actual angle differences. Additionally, a correlation penalty may be determined based on an extent of spatial correlation in the ridge flow angle differences in the difference map.

Claims

1. An electronic device, comprising:

a fingerprint recognition device for capturing a fingerprint image; and

a processing device operatively connected to the fingerprint recognition device and adapted to raster scan a ridge flow map associated with the captured fingerprint image and determine one or more optimal weights for the ridge flow map, wherein;

the one or more optimal weights are based on a prediction of a ridge flow angle for each cell in the ridge flow map using an actual ridge flow angle in one or more previously-scanned neighboring cells; and

the processing device is adapted to perform an autoregressive modeling of the actual ridge flow angles in the ridge flow map.

2. The electronic device as in claim 1, further comprising a memory operatively connected to the processing device for storing one or more ridge flow maps.

3. The electronic device as in claim 1, wherein the processing device is adapted to determine the one or more optimal weights for the ridge flow map based on minimizing, for each ridge flow in at least a portion of the ridge flow map, an error between an actual angle of a ridge flow and a predicted angle of the ridge flow.

4. The electronic device as in claim 3, wherein the processing device is adapted to determine the one or more optimal weights for the at least a portion of the ridge flow map to minimize a least squares error between the actual and predicted angles of the ridge flow.

5. The electronic device as in claim 1, wherein, to determine the one or more optimal weights for the ridge flow map associated with the captured fingerprint image, the processing device is adapted to:

determine a residual map based on differences between each actual and predicted angle in the ridge flow map using the one or more optimal weights; and

determine entropy of the residual map.

6. A method for processing a fingerprint captured by a fingerprint recognition device, the method comprising:

generating a ridge flow map based on a captured fingerprint image;

raster scanning the ridge flow map;

determining one or more optimal weights for the ridge flow map based on predicting a ridge flow angle for each cell in the ridge flow map using an actual ridge flow angle in one or more previously-scanned neighboring cells; and

determining a residual map based on differences between each actual and predicted angle in the ridge flow map using the one or more optimal weights; and

determining entropy of the residual map.

7. The method as in claim 6, wherein the one or more previously-scanned neighboring cells comprise at least one neighboring cell in a same row.

8. The method as in claim 6, wherein the one or more previously-scanned neighboring cells comprise at least one neighboring cell in an immediately preceding row.

9. The method as in claim 6, wherein the one or more optimal weights are determined so as to minimize error between the predicted and actual ridge flow angles.

10. A method for processing a fingerprint captured by a fingerprint recognition device, the method comprising:

aligning a captured ridge flow map with an enrolled ridge flow map, wherein the captured ridge flow map is based on the captured fingerprint;

estimating entropy for the captured ridge flow map, wherein estimating the entropy comprises;

determining one or more optimal weights for the captured ridge flow map by;

raster scanning the captured ridge flow map;

predicting an angle in each cell of the captured ridge flow map based on an actual ridge flow angle in one or more previously-scanned neighboring cells;

determining a first residual map based on differences between each actual and predicted angle of the captured ridge flow map using the one or more optimal weights; and

determining entropy of the first residual map; and

estimating entropy for the enrolled ridge flow map, wherein estimating the entropy comprises;

determining one or more optimal weights for the enrolled ridge flow map by;

raster scanning the enrolled ridge flow map;

predicting an angle in each cell of the enrolled ridge flow map based on an actual ridge flow angle in one or more previously-scanned neighboring cells;

determining a second residual map based on differences between each actual and predicted angle using the one or more optimal weights; and

determining entropy of the second residual map.

11. The method as in claim 10, further comprising:

generating a difference map based on the enrolled and captured ridge flow maps; and

determining a match score that estimates a degree of similarity between the captured and enrolled ridge flow maps.

12. The method as in claim 11, wherein the difference map comprises angle differences between the enrolled ridge flow map and the captured ridge flow map.

13. The method as in claim 10, wherein the one or more optimal weights for the captured ridge flow map minimize errors between actual angles and predicted angles of the captured ridge flow map.

14. The method as in claim 13, wherein the one or more optimal weights for the captured ridge flow map minimize a least squares error between the actual and predicted angles.

15. The method as in claim 10, wherein the one or more previously-scanned neighboring cells in the enrolled ridge flow map and in the captured ridge flow map comprise at least one neighboring cell in a same row.

16. The method as in claim 10, wherein the one or more previously-scanned neighboring cells in the enrolled ridge flow map and in the captured ridge flow map comprise at least one neighboring cell in an immediately preceding row.

17. The method as in claim 10, wherein predicting an angle in each cell of the captured ridge flow map based on an actual ridge flow angle in one or more previously-scanned neighboring cells comprises performing an autoregressive modeling of the actual ridge flow angles in the captured ridge flow map.