Face authentication anti-spoofing using ultrasound
Techniques and apparatuses are described that implement face authentication anti-spoofing using ultrasound. In particular, a face-authentication system uses ultrasound to distinguish between a real human face and a presentation attack that uses instruments to present a version of a human face. The face-authentication system includes or communicates with an ultrasonic sensor, which can detect a presentation attack and notify the face-authentication system. In general, the ultrasonic sensor analyzes characteristics of a presented object and determines whether the object represents a human face or a presentation attack instrument. In this way, the ultrasonic sensor can prevent unauthorized actors from using the presentation attack to gain access to a user's account or information.
1. A method performed by an ultrasonic sensor, the method comprising:
transmitting an ultrasonic transmit signal;
receiving an ultrasonic receive signal, the ultrasonic receive signal comprising a version of the ultrasonic transmit signal that is reflected by an object;
generating range-profile data based on the ultrasonic receive signal, the range-profile data comprising amplitude information across a range dimension and a time dimension;
generating normalized range-profile data by normalizing the amplitude information within the range-profile data based on a determined distance to the object;
determining a feature associated with the amplitude information of the normalized range-profile data within a region-of-interest that is associated with a location of the object;
detecting a presentation attack based on a comparison of the determined feature to a threshold, the presentation attack attempting to spoof a face-authentication system, the object associated with the presentation attack; and
preventing the face-authentication system from authenticating the presentation attack.
2. The method of claim 1 , wherein:
the determining of the feature comprises detecting a peak amplitude within the region-of-interest of the range-profile data; and
the detecting the presentation attack comprises:
comparing the peak amplitude to the threshold; and
responsive to the peak amplitude being above the threshold, detecting the presentation attack.
3. The method of claim 2 , further comprising:
transmitting another ultrasonic transmit signal;
receiving another ultrasonic receive signal, the other ultrasonic receive signal comprising a version of the other ultrasonic transmit signal that is reflected by a human face;
generating other range-profile data based on the other ultrasonic receive signal, the other range-profile data comprising amplitude associated with the other ultrasonic receive signal;
detecting another peak amplitude within another region-of-interest of the other range-profile data, the other region-of-interest associated with a location of the human face;
comparing the other peak amplitude to the threshold; and
responsive to the other peak amplitude being less than the threshold, enabling the face-authentication system to authenticate the human face.
4. The method of claim 1 , wherein:
the determining of the feature comprises determining a distribution of energy over the region-of-interest; and
the detecting the presentation attack comprises detecting the presentation attack responsive to the distribution of energy being greater than the threshold.
5. The method of claim 1 , further comprising:
determining the location of the object,
wherein the region-of-interest includes the location of the object.
6. The method of claim 5 , further comprising:
accepting sensor data from a sensor,
wherein the determining the location of the object comprises determining the location of the object based on the sensor data.
7. The method of claim 6 , wherein the sensor comprises:
a phase-difference sensor of a camera; or
a radio-frequency sensor.
8. The method of claim 5 , further comprising:
prior to determining the feature, filtering the range-profile data to extract the amplitude information associated with the region-of-interest.
9. The method of claim 1 , wherein the transmitting of the ultrasonic transmit signal comprises generating the ultrasonic transmit signal having a bandwidth that results in a range resolution of the range-profile data being less than approximately 7 centimeters.
10. The method of claim 1 , wherein the normalizing of the amplitude information further comprises normalizing the amplitude information based on a cross-coupling factor.
11. The method of claim 1 , wherein the normalizing of the amplitude information further comprises normalizing the amplitude information based on a constant-false-alarm-rate signal-to-noise ratio.
12. The method of claim 1 , further comprising:
accepting motion data from an inertial measurement unit, the motion data representing motion of the ultrasonic sensor; and
compensating for the motion of the ultrasonic sensor by modifying the range-profile data based on the motion data.
13. The method of claim 1 , further comprising:
accepting motion data from an inertial measurement unit; and
determining that the ultrasonic sensor is substantially stationary based on the motion data,
wherein the transmitting the ultrasonic transmit signal comprises transmitting the ultrasonic transmit signal responsive to determining that the ultrasonic sensor is substantially stationary.
14. An apparatus comprising an ultrasonic sensor configured to:
transmit an ultrasonic transmit signal;
receive an ultrasonic receive signal, the ultrasonic receive signal comprising a version of the ultrasonic transmit signal that is reflected by an object;
generate range-profile data based on the ultrasonic receive signal, the range-profile data comprising amplitude information across a range dimension and a time dimension;
generate normalized range-profile data by normalizing the amplitude information within the range-profile data based on a determined distance to the object;
determine a feature associated with the amplitude information of the normalized range-profile data within a region-of-interest that is associated with a location of the object;
detect a presentation attack based on a comparison of the determined feature to a threshold, the presentation attack attempting to spoof a face-authentication system, the object associated with the presentation attack; and
prevent the face-authentication system from authenticating the presentation attack.
15. The apparatus of claim 14 , wherein the ultrasonic sensor is configured to:
detect a peak amplitude within the region-of-interest of the range-profile data;
compare the peak amplitude to the threshold; and
detect the presentation attack responsive to the peak amplitude being above the threshold.
16. The apparatus of claim 14 , wherein the ultrasonic sensor is configured to:
determine a distribution of energy over the region-of-interest; and
detect the presentation attack responsive to the distribution of energy being greater than the threshold.
17. The apparatus of claim 14 , wherein the ultrasonic sensor is configured to generate the ultrasonic transmit signal having a bandwidth that results in a range resolution of the range-profile data being less than approximately 7 centimeters.
18. The apparatus of claim 14 , wherein the ultrasonic sensor is further configured to normalize the amplitude information within the range-profile data based a cross-coupling factor or a constant-false-alarm-rate signal-to-noise ratio.
19. A non-transitory computer-readable medium comprising instructions which, when executed by a processor, cause an ultrasonic sensor to:
transmit an ultrasonic transmit signal;
receive an ultrasonic receive signal, the ultrasonic receive signal comprising a version of the ultrasonic transmit signal that is reflected by an object;
generate range-profile data based on the ultrasonic receive signal, the range-profile data comprising amplitude information across a range dimension and a time dimension;
generate normalized range-profile data by normalizing the amplitude information within the range-profile data based on a determined distance to the object;
determine a feature associated with the amplitude information of the normalized range-profile data within a region-of-interest that is associated with a location of the object;
detect a presentation attack based on a comparison of the determined feature to a threshold, the presentation attack attempting to spoof a face-authentication system, the object associated with the presentation attack; and
prevent the face-authentication system from authenticating the presentation attack.
20. The computer-readable medium of claim 19 , wherein the instructions cause the ultrasonic sensor to:
detect a peak amplitude within the region-of-interest of the range-profile data;
compare the peak amplitude to the threshold; and
detect the presentation attack responsive to the peak amplitude being above the threshold.