Spoof detection using dual-band near-infrared (NIR) imaging
This document described a method that includes illuminating a first subject with electromagnetic radiation in a first wavelength range, and capturing a first image of the first subject under illumination of the first wavelength range. The method also includes illuminating the first subject with electromagnetic radiation in a second wavelength range, wherein a difference between (i) absorption of the second wavelength range by live human tissue, and (ii) absorption of the first wavelength range by live human tissue satisfies a threshold condition. The method further includes capturing a second image of the first subject under illumination of the second wavelength range, identifying, based on a difference between the first image and the second image, the first subject as an alternative representation of a live person, and preventing access to the secure system responsive to identifying the first subject as an alternative representation of a live person.
1. A method for preventing access to a secure system based on determining that a subject is an alternative representation of a live person, the method comprising:
illuminating a first subject with electromagnetic radiation in a first wavelength range;
capturing a first image of the first subject under illumination of the first wavelength range;
illuminating the first subject with electromagnetic radiation in a second wavelength range, wherein a difference between (i) absorption of the second wavelength range by live human tissue, and (ii) absorption of the first wavelength range by live human tissue satisfies a threshold condition;
capturing a second image of the first subject under illumination of the second wavelength range;
calculating pixel-wise differences in intensity values between corresponding pixels of the first image and the second image;
calculating an aggregated measure of the pixel-wise differences in intensity values;
calculating a difference between the aggregated measure of the pixel-wise differences and a target value; and
identifying, based at least on the difference between the aggregated measure and the target value, the first subject as an alternative representation of a live person; and
preventing access to the secure system responsive to identifying the first subject as an alternative representation of a live person.
2. The method of claim 1 , wherein the alternative representation of a live person comprises a photograph of the live person.
3. The method of claim 2 , wherein the photograph is printed on paper or presented on a display device.
4. The method of claim 1 , wherein the first wavelength range includes 940 nm, and the second wavelength range includes 850 nm.
5. The method of claim 1 , wherein the threshold condition is satisfied if a variation between (i) the absorption of the second wavelength range by live human tissue, and (ii) the absorption of the first wavelength range by live human tissue, is more than 40%.
6. The method of claim 1 , wherein the aggregated measure of the pixel-wise differences in intensity values comprises an average of the pixel-wise differences in intensity values.
7. The method of claim 1 , wherein identifying the first subject as the alternative representation comprises processing the first image and the second image using a machine learning process trained to discriminate between images of live persons and corresponding alternative representations of the live persons.
8. The method of claim 1 , further comprising:
illuminating a second subject with electromagnetic radiation using the first wavelength range;
capturing a third image of the second subject under illumination of the first wavelength range;
illuminating the second subject with electromagnetic radiation using the second wavelength range;
capturing a fourth image of the second subject under illumination of the second wavelength range;
identifying, based on a difference between the third image and the fourth image, the second subject as a live person; and
responsive to identifying the second subject as a live person, initiating an authentication process for determining if the live person is authorized to access the secure system.
9. A secure system comprising:
one or more illumination sources that illuminate a first subject with electromagnetic radiation in a first wavelength range during a first time period, and with electromagnetic radiation in a second wavelength range during a second time period, wherein a difference between (i) absorption of the second wavelength range by live human tissue, and (ii) absorption of the first wavelength range by live human tissue satisfies a threshold condition;
one or more image acquisition devices that capture a first image of the first subject under illumination of the first wavelength range, and a second image of the first subject under illumination of the second wavelength range; and
an image analysis engine comprising one or more processing devices, the image analysis engine programmed to:
calculate pixel-wise differences in intensity values between corresponding pixels of the first image and the second image,
calculate an aggregated measure of the pixel-wise differences in intensity values,
calculate a difference between the aggregated measure of the pixel-wise differences and a target value,
identify, based at least on the difference between the aggregated measure and the target value, the first subject as an alternative representation of a live person, and
prevent access to the secure system responsive to identifying the first subject as an alternative representation of a live person.
10. The system of claim 9 , wherein the alternative representation of a live person comprises a photograph of the live person.
11. The system of claim 10 , wherein the photograph is printed on paper, or presented on a display device.
12. The system of claim 9 , wherein the first wavelength range includes 940 nm, and the second wavelength range includes 850 nm.
13. The system of claim 9 , wherein the threshold condition is satisfied if a variation between (i) the absorption of the second wavelength range by live human tissue, and (ii) the absorption of the first wavelength range by live human tissue, is more than 40%.
14. The system of claim 9 , wherein the aggregated measure of the pixel-wise differences in intensity values comprises an average of the pixel-wise differences in intensity values.
15. The system of claim 9 , wherein identifying the first subject as the alternative representation comprises processing the first image and the second image using a machine learning process trained to discriminate between images of live persons and corresponding alternative representations of the live persons.
16. The system of claim 9 , wherein:
the one or more illuminating sources
illuminate a second subject with electromagnetic radiation in the first wavelength range, and
illuminate the second subject with electromagnetic radiation in the second wavelength range;
the one or more image acquisition devices
capture a third image of the second subject under illumination of the first wavelength range, and
capture a fourth image of the second subject under illumination of the second wavelength range; and
the image analysis engine is programmed to:
identify, based on a difference between the third image and the fourth image, the second subject as a live person, and
responsive to identifying the second subject as a live person, initiate an authentication process for determining if the live person is authorized to access the secure system.
17. The system of claim 9 , wherein the secure system comprises a user-interface disposed in a kiosk.
18. One or more non-transitory machine-readable storage devices encoded with instructions configured to cause one or more processing devices to execute operations comprising:
causing one or more light source arrays to illuminate a first subject with electromagnetic radiation in a first wavelength range;
receiving a first image of the first subject captured under illumination of the first wavelength range;
causing the one or more light source arrays to illuminate the first subject with electromagnetic radiation in a second wavelength range, wherein a difference between (i) absorption of the second wavelength range by live human tissue, and (ii) absorption of the first wavelength range by live human tissue satisfies a threshold condition;
receiving a second image of the first subject captured under illumination of the second wavelength range;
calculating pixel-wise differences in intensity values between corresponding pixels of the first image and the second image;
calculating an aggregated measure of the pixel-wise differences in intensity values;
calculating a difference between the aggregated measure of the pixel-wise differences and a target value; and
identifying, based at least on the difference between the aggregated measure and the target value, the first subject as an alternative representation of a live person; and
preventing access to a secure system responsive to identifying the first subject as an alternative representation of a live person.