Systems and methods for coarse-to-fine ridge-based biometric image alignment

Patent metadata

Field	Value
Publication number	US-9792485-B2
Application number	US-201514788662-A
Country	US
Kind code	B2
Filing date	Jun 30, 2015
Priority date	Jun 30, 2015
Publication date	Oct 17, 2017
Grant date	Oct 17, 2017

Abstract

Systems and methods for aligning images are disclosed. A method includes: receiving a first skeletonized biometric image; generating a first coarse representation of the first skeletonized biometric image; identifying a set of candidate transformations that align the first skeletonized biometric image to a second skeletonized biometric image based on comparing transformed versions of the first coarse representation to a second coarse representation of the second skeletonized biometric image; selecting a first candidate transformation as the candidate transformation that minimizes a difference metric between a transformed version of the first skeletonized biometric image and the second skeletonized biometric image; and determining whether the first skeletonized biometric image transformed by the first candidate transformation matches the second skeletonized biometric image, wherein the first skeletonized biometric image transformed by the first candidate transformation matches the second skeletonized biometric image if the difference metric satisfies a threshold.

First claim

What is claimed is:

1. A processing system, comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the processing system to align a first image to a second image, wherein the first and second images are biometric images, by performing the steps of:
generating a first orientation map corresponding to the first image, wherein the first orientation map is an image in which each pixel of the image represents a local orientation of an area of the first image covered by the pixel;
determining a set of candidate transformations that, when applied to the first orientation map, align a transformed version of the first orientation map to a second orientation map corresponding to the second image;
for each candidate transformation in the set of candidate transformations:
applying the candidate transformation to the first image to generate a transformed first image, and
calculating a difference metric between the transformed first image and the second image; and

selecting a first candidate transformation from the set of candidate transformations that, when applied to the first image, minimizes the difference metric between the transformed first image and the second image.

2. The processing system of claim 1 , wherein the steps further include:
determining whether the first image transformed by the first candidate transformation matches the second image, wherein the first image transformed by the first candidate transformation matches the second image if the difference metric satisfies a threshold.

3. The processing system of claim 2 , wherein the first and second images are skeletonized biometric images, and determining that the first image transformed by the first candidate transformation matches the second image comprises determining a fingerprint match based on a difference between contours in the skeletonized biometric images.

4. The processing system of claim 1 , wherein each of the candidate transformations comprises a translation value and a rotation value.

5. The processing system of claim 1 , wherein determining the set of candidate transformations comprises:
for each transformation in a set of possible transformations that can be applied to the first orientation map, applying the transformation to the first orientation map and comparing the transformed version of the first orientation map to the second orientation map; and
determining the set of candidate transformations based on selecting a number of transformations from the set of possible transformations that result in a minimum difference between the transformed version of the first orientation map and the second orientation map.

6. The processing system of claim 1 , wherein the steps further include:
sampling the first orientation map to generate a first coarse orientation map;
sampling the second orientation map to generate a second coarse orientation map; and
determining a set of coarse candidate transformations to align the first coarse orientation map to the second coarse orientation map;
wherein determining the set of candidate transformations comprises:
for each coarse candidate transformation in the set of coarse candidate transformations:
applying the coarse candidate transformation to the first orientation map to generate a transformed first orientation map,
calculating a difference metric between the transformed first orientation map and the second orientation map, and
determining the set of candidate transformations based on selecting a number of coarse candidate transformations that result in a minimum difference between the transformed first orientation map and the second orientation map.

7. The processing system of claim 6 , wherein determining the set of coarse candidate transformations to align the first coarse orientation map to the second coarse orientation map comprises:
for each transformation in a set of possible transformations that can be applied to the first coarse orientation map, applying the transformation to the first coarse orientation map and comparing a transformed version of the first coarse orientation map to the second coarse orientation map; and
identifying an initial set of hypothesis transformations based on similarity metrics resulting from applying each of the transformations in the set of possible transformations to the first coarse orientation map and comparing the transformed version of the first coarse orientation map to the second coarse orientation map; and
identifying additional hypothesis transformations to add to the initial set of hypothesis transformations based on the additional hypothesis transformations being within a threshold of at least one hypothesis transformation in the initial set of hypothesis transformations.

8. The processing system of claim 1 , wherein selecting the first candidate transformation from the set of candidate transformations that minimizes the difference metric between the transformed first image and the second image comprises:
for each candidate transformation, performing an iterative closest point (ICP) operation to minimize a difference between the first image transformed by the candidate transformation and the second image.

9. The processing system of claim 1 , wherein the first orientation map has the same dimensions as the first image.

10. The processing system of claim 1 , wherein the first orientation map has dimensions that are smaller than dimensions of the first image.

11. The processing system of claim 1 , wherein the first orientation map is generated based on a grayscale image corresponding to the first image.

12. The processing system of claim 1 , wherein the first image represents a portion of a fingerprint.

13. A method, comprising:
receiving a first skeletonized biometric image;
generating a first coarse representation of the first skeletonized biometric image, wherein the first coarse representation is an image in which each pixel of the image represents a local orientation of an area of the first skeletonized biometric image covered by the pixel;
identifying a set of candidate transformations that align the first skeletonized biometric image to a second skeletonized biometric image based on comparing transformed versions of the first coarse representation to a second coarse representation of the second skeletonized biometric image;
selecting a first candidate transformation from the set of candidate transformations as the candidate transformation that, when applied to the first skeletonized biometric image, minimizes a difference metric between a transformed version of the first skeletonized biometric image and the second skeletonized biometric image; and
determining whether the first skeletonized biometric image transformed by the first candidate transformation matches the second skeletonized biometric image, wherein the first skeletonized biometric image transformed by the first candidate transformation matches the second skeletonized biometric image if the difference metric satisfies a threshold.

14. The method of claim 13 , wherein the first coarse representation comprises an orientation map of the first skeletonized biometric image.

15. The method of claim 13 , wherein the first coarse representation comprises a density map of the first skeletonized biometric image.

16. The method of claim 13 , wherein the first skeletonized biometric image represents a portion of a fingerprint.

17. A device, comprising:
a processing system configured to:
receive a first skeletonized biometric image;
generate a first coarse representation of the first sk

Assignees

• Synaptics Inc

Inventors

• Oreifej Omar
• Sengupta Kuntal
• Schwartz Adam
• Chinni Krishna

Classifications

• G06V40/1376

  Matching features related to ridge properties or fingerprint texture · CPC title

• G06V40/1318 Primary

  using electro-optical elements or layers, e.g. electroluminescent sensing · CPC title

• G06V10/751

  Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching · CPC title

• G06K9/001 Primary

  Physics · mapped topic

• G06K9/6202

  Physics · mapped topic