FEATURE EXTRACTION AND MATCHING FOR BIOMETRIC AUTHENTICATION

• US 20150286864A1
• Filed: 05/09/2014
• Published: 10/08/2015
• Est. Priority Date: 09/16/2013
• Status: Active Grant

• Current Assignee: Jumio Inc.
• Original Assignee: EyeVerify, Inc. (Ant Group Co., Ltd.)
• Inventors: Gottemukkula, Vikas, Derakhshani, Reza R., Saripalle, Sashi K.
• Granted Patent: US 9,390,327 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06T 11/60 Editing figures and text; Combining figures or text
• CPC Class Codes: G06T 3/00 Geometric image transformations in the plane of the image
• CPC Class Codes: G06V 10/467 Encoded features or binary features, e.g. local binary patterns [LBP]
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
• CPC Class Codes: G06V 40/14 Vascular patterns
• CPC Class Codes: G06V 40/18 Eye characteristics, e.g. of the iris
• CPC Class Codes: G06V 40/19 Sensors therefor
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction
• CPC Class Codes: G06V 40/197 Matching; Classification
• CPC Class Codes: G06V 40/50 Maintenance of biometric data or enrolment thereof

Abstract

In a feature extraction and pattern matching system, image sharpening can enable vascular point detection (VPD) for detecting points of interest from visible vasculature of the eye. Pattern Histograms of Extended Multi-Radii Local Binary Patterns and/or Pattern Histograms of Extended Multi-Radii Center Symmetric Local Binary Patterns can provide description of portions of images surrounding a point of interest, and enrollment and verification templates can be generated using points detected via VPD and the corresponding descriptors. Inlier point pairs can be selected from the enrollment and verification templates, and a first match score indicating similarity of the two templates can be computed based on the number of inlier point pairs and one or more parameters of a transform selected by the inlier detection. A second match score can be computed by applying the selected transform, and either or both scores can be used to authenticate the user.

Claims

1. A computer-implemented method comprising:

obtaining a sharpened image based on a plurality of images, at least one image in the plurality comprising an image of a vascular structure;

detecting a plurality of vascular points in the sharpened image;

for each of a plurality of the detected vascular points, generating a respective plurality of different local image descriptors, wherein generating the respective plurality of local image descriptors comprises computing at least one of;

(i) a pattern histogram of extended multi-radii local binary patterns (PH-EMR-LBP) of an image region surrounding the detected vascular point, and (ii) a pattern histogram of extended multi-radii center-symmetric local binary patterns (PH-EMR-CS-LBP) of an image region surrounding the detected vascular point, computing both the PH-EMR-LBP and PH-EMR-CS-LBP comprising tiling a selected neighborhood region into overlapping sub regions; and

generating a template comprising a plurality of the vascular points and their respective local image descriptors.

3. The method of claim 1, wherein obtaining a particular sharpened image comprises:

selecting a set of images from the plurality of images and averaging the images in the set to generate an average image;

convolving the average image with a plurality of Gabor kernels oriented at different angles, to obtain the particular sharpened image.

4. The method of claim 1, wherein detecting the plurality of vascular points comprises:

selecting a point in the sharpened image as a candidate vascular point;

identifying a first plurality of N neighborhood points according to first window centered on the candidate vascular point and selecting a second plurality of N neighborhood points according to second, different window centered on the candidate vascular point;

determining a plurality of N states comprising a state for each neighborhood point in the first plurality of N points, by performing a comparison based on, at least in part, respective intensities of the point in the first plurality of N points, a corresponding point in the second plurality of N points, and the candidate vascular point;

aggregating the N states; and

designating the candidate vascular point as a vascular point based on, at least in part, the aggregated states.

5. The method of claim 4, wherein:

the comparison comprises;

(i) testing if an intensity of the point in the first plurality of N points is greater by a first threshold than an intensity of the candidate vascular point, and (ii) testing if an intensity of the corresponding point in the second plurality of N points is greater by the first threshold than the intensity of the candidate vascular point; and

determining a corresponding state comprises setting the state to a first value if any of the tests (i) and (ii) is true, and setting the state to a second value otherwise.

6. The method of claim 4, wherein:

aggregating the N states comprises summing the N states; and

designating comprises testing if a summed value of the aggregated states exceeds a selected count;

7. The method claim 4, wherein:

the comparison comprises testing at least one of whether;

(a) the intensity of the point in the first plurality of N points is greater than a second threshold, and (b) the intensity of the corresponding point in the second plurality of N points is greater than the second threshold; and

determining the corresponding state comprises setting the state to a second value if any of the tests (a) and (b) is true.

8. The method of claim 4, further comprising:

performing the selecting, identifying, determining, aggregating, and designating for a plurality of points in the sharpened image;

setting each candidate vascular point designated as a vascular point to a first value and setting other candidate vascular points to a second value to obtain a binary vascular map (BVM) representing veins.

9. The method of claim 8, further comprising thinning the BVM by excluding at least one vascular point that:

(i) corresponds to a boundary, and (ii) was set to the first value.

10. The method of claim 4, further comprising locally suppressing at least one vascular point that was set to the first value, the local suppression being based on, at least in part, a gradient magnitude map relating to at least a portion of the sharpened image.

11. The method of claim 1, wherein generating the respective plurality of local image descriptors comprises computing at least one of:

(i) a pattern histogram of extended multi-radii local binary patterns of an image region surrounding the detected vascular point, and (ii) a pattern histogram of extended multi-radii center-symmetric local binary patterns of an image region surrounding the detected vascular point.

12. The method of claim 1, further comprising shortening one of the plurality local image descriptors using principal component analysis (PCA), to eliminate dimensions that do not contribute to local descriptor variances.

30. The method of claim 1, wherein obtaining a particular sharpened image comprises:

selecting a set of images from the plurality of images and averaging the images in the set to generate an average image;

convolving the average image with a Laplacian of Gaussian (LoG) kernel, to obtain a convolved image;

subtracting each pixel of the convolved image from a maximum pixel value, to obtain a difference image; and

multiplying, in a pixel-wise manner, the difference image and the average image, to obtain the particular sharpened image.

2. (canceled)

13-29. -29. (canceled)

31. A system comprising:

a memory having instructions stored thereon; and

a processor programmed to execute the instructions to perform operations comprising;

obtaining a sharpened image based on a plurality of images, at least one image in the plurality comprising an image of a vascular structure;

detecting a plurality of vascular points in the sharpened image;

for each of a plurality of the detected vascular points, generating a respective plurality of different local image descriptors, wherein generating the respective plurality of local image descriptors comprises computing at least one of;

(i) a pattern histogram of extended multi-radii local binary patterns (PH-EMR-LBP) of an image region surrounding the detected vascular point, and (ii) a pattern histogram of extended multi-radii center-symmetric local binary patterns (PH-EMR-CS-LBP) of an image region surrounding the detected vascular point, computing both the PH-EMR-LBP and PH-EMR-CS-LBP comprising tiling a selected neighborhood region into overlapping sub regions; and

generating a template comprising a plurality of the vascular points and their respective local image descriptors.

32. The system of claim 31, wherein for obtaining a particular sharpened image, the processor is further programmed to perform operations comprising:

selecting a set of images from the plurality of images and averaging the images in the set to generate an average image;

convolving the average image with a Laplacian of Gaussian (LoG) kernel, to obtain a convolved image;

subtracting each pixel of the convolved image from a maximum pixel value, to obtain a difference image; and

multiplying, in a pixel-wise manner, the difference image and the average image, to obtain the particular sharpened image.

33. The system of claim 31, wherein for obtaining a particular sharpened image, the processor is further programmed to perform operations comprising:

selecting a set of images from the plurality of images and averaging the images in the set to generate an average image;

convolving the average image with a plurality of Gabor kernels oriented at different angles, to obtain the particular sharpened image.

34. The system of claim 31, wherein for detecting the plurality of vascular points, the processor is further programmed to perform operations comprising:

selecting a point in the sharpened image as a candidate vascular point;

identifying a first plurality of N neighborhood points according to first window centered on the candidate vascular point and selecting a second plurality of N neighborhood points according to second, different window centered on the candidate vascular point;

determining a plurality of N states comprising a state for each neighborhood point in the first plurality of N points, by performing a comparison based on, at least in part, respective intensities of the point in the first plurality of N points, a corresponding point in the second plurality of N points, and the candidate vascular point;

aggregating the N states; and

designating the candidate vascular point as a vascular point based on, at least in part, the aggregated states.

35. The system of claim 34, wherein:

to perform the comparison the processor is further programmed to perform operations comprising;

(i) testing if an intensity of the point in the first plurality of N points is greater by a first threshold than an intensity of the candidate vascular point, and (ii) testing if an intensity of the corresponding point in the second plurality of N points is greater by the first threshold than the intensity of the candidate vascular point; and

for determining a corresponding state the processor is further programmed to perform operations comprising setting the state to a first value if any of the tests (i) and (ii) is true, and setting the state to a second value otherwise.

36. The system of claim 34, wherein:

for aggregating the N states the processor is further programmed to perform operations comprising summing the N states; and

for designating the processor is further programmed to perform operations comprising testing if a summed value of the aggregated states exceeds a selected count;

37. The system claim 34, wherein:

to perform the comparison the processor is further programmed to perform operations comprising testing at least one of whether;

(a) the intensity of the point in the first plurality of N points is greater than a second threshold, and (b) the intensity of the corresponding point in the second plurality of N points is greater than the second threshold; and

for determining the corresponding state the processor is further programmed to perform operations comprising setting the state to a second value if any of the tests (a) and (b) is true.

38. The system of claim 34, wherein the processor is further programmed to perform operations comprising:

performing the selecting, identifying, determining, aggregating, and designating for a plurality of points in the sharpened image;

setting each candidate vascular point designated as a vascular point to a first value and setting other candidate vascular points to a second value to obtain a binary vascular map (BVM) representing veins.

39. The system of claim 38, wherein the processor is further programmed to perform operations comprising thinning the BVM by excluding at least one vascular point that:

(i) corresponds to a boundary, and (ii) was set to the first value.

40. The system of claim 34, wherein the processor is further programmed to perform operations comprising locally suppressing at least one vascular point that was set to the first value, the local suppression being based on, at least in part, a gradient magnitude map relating to at least a portion of the sharpened image.

41. The system of claim 31, wherein for generating the respective plurality of local image descriptors the processor is further programmed to perform operations comprising computing at least one of:

(i) a pattern histogram of extended multi-radii local binary patterns of an image region surrounding the detected vascular point, and (ii) a pattern histogram of extended multi-radii center-symmetric local binary patterns of an image region surrounding the detected vascular point.

42. The system of claim 31, wherein the processor is further programmed to perform operations comprising shortening one of the plurality local image descriptors using principal component analysis (PCA), to eliminate dimensions that do not contribute to local descriptor variances.