Iris recognition via plenoptic imaging
Iris recognition can be accomplished for a wide variety of eye images by using plenoptic imaging. Using plenoptic technology, it is possible to correct focus after image acquisition. One example technology reconstructs images having different focus depths and stitches them together, resulting in a fully focused image, even in an off-angle gaze scenario. Another example technology determines three-dimensional data for an eye and incorporates it into an eye model used for iris recognition processing. Another example technology detects contact lenses. Application of the technologies can result in improved iris recognition under a wide variety of scenarios.
1. A method implemented at least in part by a computing system, the method comprising:
receiving a single plenoptic off-angle iris image depicting an iris with an off-angle gaze;
from the single plenoptic off-angle iris image depicting an iris with an off-angle gaze, creating a plurality of reconstructed images at different synthetic focal depths, wherein the plurality of reconstructed images comprise different in-focus regions of the iris, and the different synthetic focal depths for the plurality of reconstructed images are chosen using one or more selected from the group consisting of an amount of entropy in the reconstructed images, the frequency response of the lines in the reconstructed images, and the contrast between neighboring pixels of the reconstructed images;
stitching together the different in-focus regions of the iris with the off-angle gaze from the plurality of reconstructed images, wherein the stitching generates a fully focused image of the iris with the off-angle gaze originating from the single plenoptic off-angle iris image;
constructing a reconstructed synthetic eye from the fully focused image of the iris with the off-angle gaze; and
generating a frontal view of the iris from the reconstructed synthetic eye.
2. The method of claim 1 further comprising:
submitting the fully focused image of the iris to an iris recognizer for iris recognition processing.
3. The method of claim 2 wherein:
the iris recognition processing comprises transforming, masking, and machine learning.
4. The method of claim 1 further comprising:
dividing a particular reconstructed image into a plurality of regions;
determining that at least one of the plurality of regions is in focus; and
using the at least one in-focus region for the stitching together.
5. The method of claim 1 further comprising:
dividing the single plenoptic off-angle iris image into a plurality of overall regions;
identifying respective of the reconstructed images having in-focus regions corresponding to the overall regions; and
wherein the stitching together stitches together the in-focus regions corresponding to the overall regions.
6. The method of claim 1 further comprising:
identifying the different in-focus regions in respective of the reconstructed images.
7. The method of claim 6 wherein the identifying comprises analyzing entropy.
8. The method of claim 6 wherein the identifying comprises analyzing contrast between neighboring pixels.
9. The method of claim 1 wherein:
the different in-focus regions of the iris are generated by applying different synthetic focus planes to the single plenoptic off-angle iris image.
10. The method of claim 1 wherein:
gaze angle of the off-angle gaze is at or more than 50 degrees away from frontal.
11. The method of claim 1 further comprising:
determining a portion of the iris that is occluded by a limbus in an on-angle gaze; and
removing the occluded portion of the iris from consideration during iris recognition.
12. One or more computer-readable media having encoded thereon computer-executable instructions causing a computing system to perform a method comprising:
receiving a single plenoptic off-angle iris image depicting an iris with an off-angle gaze;
from the single plenoptic off-angle iris image depicting an iris with an off-angle gaze, creating a plurality of reconstructed images at different synthetic focal depths, wherein the plurality of reconstructed images comprise different in-focus regions of the iris and are generated by applying different synthetic focus planes to the single plenoptic off-angle iris image, wherein refocusing is performed after the single plenoptic off-angle iris image is captured; and
stitching together the different in-focus regions of the iris with the off-angle gaze from the plurality of reconstructed images, wherein the stitching generates a fully focused image of the iris with the off-angle gaze originating from the single plenoptic off-angle iris image;
constructing a reconstructed synthetic eye from the fully focused image of the iris with the off-angle gaze;
generating a frontal view of the iris from the reconstructed synthetic eye;
determining a portion of the iris that is occluded by a limbus in an on-angle gaze; and
removing the occluded portion of the iris from consideration during iris recognition.
13. The one or more computer-readable media of claim 12 wherein:
gaze angle of the off-angle gaze is at or more than 50 degrees away from frontal.
14. The one or more computer-readable media of claim 12 wherein the method further comprises:
submitting the fully focused image of the iris to an iris recognizer for iris recognition processing.
15. The one or more computer-readable media of claim 14 wherein:
the iris recognition processing comprises transforming, masking, and machine learning.
16. A computing system comprising:
one or more processors;
memory; and
a fully focused image stitching engine configured to accept a plurality of reconstructed images of different synthetic focus depths of an iris with an off-angle gaze and output a fully focused image;
a frontal view reconstruction engine comprising a ray-tracing engine and eye model generating a synthetic eye, wherein the frontal view reconstruction engine accepts the fully focused image as input and outputs a frontal view of the synthetic eye;
wherein the reconstructed images are generated from a single plenoptic iris image of an iris with an off-angle gaze by applying different synthetic focus planes to the single plenoptic iris image of an iris with an off-angle gaze; and
wherein a portion of the iris that is determined as occluded by a limbus in an on-angle gaze is removed from consideration during iris recognition.
17. The computing system of claim 16 further comprising:
an iris recognition engine configured to receive the fully focused image and perform iris recognition processing.
18. The computing system of claim 17 wherein:
the iris recognition processing comprises transforming, masking, and machine learning.
19. The computing system of claim 16 wherein:
gaze angle of the off-angle gaze is at or more than 50 degrees away from frontal.