Pattern recognition for face authentication spoof prevention

Abstract

Techniques and apparatuses are described that implement pattern recognition for face authentication anti-spoofing. In particular, a face authentication system distinguishes between a real face and a presentation attack that presents a virtual face using a display. The operational settings of the camera system of the face authentication system are tailored such that a pattern associated with the operation of the display can be detected. A spoof detector of the face authentication system analyzes one or more images captured by the camera system and determines whether the pattern is presented within the images. If the pattern is presented, the face authentication system does not provide face authentication. Alternatively, if the pattern is not presented, the face authentication system can provide face authentication if face recognition is successful. In this way, the face authentication system can prevent an unauthorized participant from gaining access to a user's account or information using a presentation attack.

Description

BACKGROUND

[0001] Face authentication provides a convenient way for users to unlock their devices, increase security for accessing accounts, or sign transactions, which enhances user experience. Some face authentication systems rely on cameras for face authentication. However, it can be difficult for cameras to distinguish between a user’s face and an image of the user’s face. Thus, there is a challenge in preventing an unauthorized participant from spoofing a face authentication system that relies on a camera. SUMMARY

[0002] Techniques and apparatuses are described that implement pattern recognition for face authentication spoof prevention. In particular, a face authentication system distinguishes between a real face and a presentation attack that presents a virtual face (e.g., a digital picture of a face) using a display. The face authentication system includes a camera system and a spoof detector. The camera system captures one or more images for face authentication. The operational settings of the camera system of the face authentication system are tailored such that the spoof detector is able to detect a pattern associated with the operation of the display used in the presentation attack. The spoof detector analyzes the images captured by the camera system and determines whether the pattern is presented within the images. If the pattern is presented, the face authentication system does not provide face authentication. Or, if the pattern is not presented, the face authentication system can provide face authentication if face recognition is successful. In this way, the face authentication system can prevent an unauthorized participant from gaining access to a user’s account or information using a presentation attack.

[0003] Aspects described below include a method of a face authentication system that performs pattern recognition for face authentication spoof prevention. The method includes setting a capture rate of a camera system of the face authentication system to be greater than a threshold for detecting a pattern associated with an operation of a display. The method also includes capturing at least one image of a subject using the capture rate. Further, the method includes identifying the pattern within the at least one image. In response to identifying the pattern, the method includes determining that the subject includes the display. In response to determining that the subject includes the display, the method further includes preventing face authentication.

[0004] Aspects described below also include an apparatus that includes a face authentication system configured to perform any of the described methods.

[0005] Aspects described below include a computer-readable storage medium that includes computer-executable instructions that, in response to execution by a processor, cause a face authentication system to perform any of the described methods.

[0006] Aspects described below also include a system with an apparatus for performing pattern recognition for face authentication spoof prevention. BRIEF DESCRIPTION OF DRAWINGS

[0007] The following figures illustrate an apparatus and technique for implementing pattern recognition to prevent facial authentication spoofing. Throughout the figures, the same reference numerals denote the same features and components:

[0008] Figure 1 An exemplary environment is shown that can implement pattern recognition for anti-spoofing facial authentication;

[0009] Figure 2 An exemplary implementation of a face authentication system as part of a user device is shown;

[0010] Figure 3 An exemplary face authentication system is shown that performs pattern recognition for anti-spoofing purposes in face authentication;

[0011] Figure 4 An exemplary pulse width modulation pattern is shown;

[0012] Figure 5 An example black frame insert pattern is shown;

[0013] Figure 6 An exemplary backlight scanning pattern is shown;

[0014] Figure 7 An exemplary method is shown for performing pattern recognition operations for anti-spoofing facial authentication; and

[0015] Figure 8 An exemplary computing system is shown that is specifically implemented to perform a face authentication system for pattern recognition for anti-spoofing, or an exemplary computing system in which the technology enabling the use of a face authentication system for pattern recognition for anti-spoofing can be implemented. Detailed Implementation

[0016] Overview

[0017] Facial recognition provides users with a convenient way to unlock their devices, increase security for accessing accounts, or sign transactions, thus enhancing the user experience. Some facial recognition systems rely on cameras for authentication. However, for a camera, distinguishing between images of a user's face can be difficult. Therefore, challenges arise in preventing unauthorized actors from spoofing camera-based facial recognition systems.

[0018] Some technologies distinguish between real faces and images of faces by detecting blinks or facial movements. However, these technologies may require additional capture time, increasing the latency involved in performing facial authentication. Furthermore, these technologies can overcome this by using video recordings of faces that include these movements.

[0019] To address these issues, techniques and devices implementing pattern recognition for face authentication anti-spoofing are described herein. In particular, a face authentication system distinguishes between a real face and a presentation attack that presents a virtual face (e.g., a digital picture of a face) using a display. The face authentication system includes a camera system and a spoof detector. The camera system captures one or more images for face authentication. The camera system’s operational settings are tailored such that the spoof detector is able to detect a pattern associated with the operation of the display used in the presentation attack. The spoof detector analyzes the images captured by the camera system and determines whether the pattern is presented within the images. If the pattern is presented, the face authentication system does not provide face authentication. Or, if the pattern is not presented, the face authentication system can provide face authentication if the face recognition is successful. In this way, the face authentication system can prevent an unauthorized participant from using a presentation attack to gain access to a user’s account or information.

[0020] Example Environment

[0021] Figure 1 is an illustration of an example environment 100 in which techniques for face authentication anti-spoofing using pattern recognition can be implemented. In the environment 100, a user device 102 performs face authentication. In some cases, a user 104 controls the user device 102 and uses face authentication to access an application or sign a transaction. During face authentication, the user device 102 captures one or more images 106. In this case, the images 106 include the face of the user 104. Using face recognition techniques, the user device 102 identifies and authenticates the user 104.

[0022] In other cases, an unauthorized participant 108 can control the user device 102. In this case, the unauthorized participant 108 uses a second device (e.g., a spoofing device 110) to attempt to spoof (e.g., trick) the user device 102 into granting the unauthorized participant 108 access to the user’s 104 account or information. The spoofing device 110 can be a smartphone, a tablet, a wearable device, a television, or a virtual or augmented reality headset. The spoofing device 110 includes a display 114. The display 114 can be a light-emitting diode (LED) display or a liquid crystal display (LCD). Example LED displays include an organic LED (OLED) display, a polymer or plastic OLED (PLED or P-OLED) display, or a glass OLED (GLED or G-OLED) display. Liquid crystal displays can or can not use backlighting. On the display 114, the spoofing device 110 presents a digital picture 112 of the user 104.

[0023] During facial authentication, the unauthorized participant 108 orients the display 114 of the spoof device 110 toward the camera of the user device 102 so that the user device 102 captures an image 106 of the digital picture 112 presented on the display 114. This action represents a type of presentation attack 116. This type of presentation attack 116 can fool some authentication systems of the user device 102 and cause the user device 102 to grant the unauthorized participant 108 access. However, through the techniques for pattern recognition for facial authentication anti-spoofing, the user device 102 determines that the image 106 captures a portion of the display 114. With this knowledge, the user device 102 denies the unauthorized participant 108 authorization.

[0024] Through the described techniques, the user device 102 can distinguish between a human face (e.g., the face of the user 104) and a virtual human face (e.g., the digital picture 112) presented through various different types of displays 114. Specifically, the user device 102 identifies patterns associated with the operation of the display 114. These operations can include pulse width modulation that some displays use to control brightness. Other operations can include black frame insertion or backlight scanning that some displays use to reduce motion blur. Regarding Figure 2 The user device 102 is further described.

[0025] Figure 2 A facial authentication system 202 is shown as part of a user device 102. The user device 102 is illustrated through various non-limiting exemplary devices, including a desktop computer 102-1, a tablet 102-2, a laptop computer 102-3, a television 102-4, a computing watch 102-5, a home automation and control system 102-6, a gaming system 102-7, a microwave oven 102-8, and a vehicle 102-9. Other devices can also be used, such as a home service device, a smart thermostat, a security camera, a baby monitor, a Wi-Fi TM router, a drone, a trackpad, a graphics tablet, a netbook, an e-reader, a wall display, and other home appliances. Note that the user device 102 can be wearable, non-wearable but mobile, or relatively immobile (e.g., desktop computers and appliances). The facial authentication system 202 can be used as a standalone facial authentication system, with many different devices or peripherals, or embedded within, such as in a control panel that controls home appliances and systems, in a car, or as an accessory to a laptop computer.

[0026] The user device 102 includes one or more computer processors 204 and at least one computer-readable medium 206, which includes memory media and storage media. Applications and / or an operating system (not shown) embodied as computer-readable instructions on the computer-readable medium 206 can be executed by the computer processors 204 to provide some of the functionality described herein. The computer-readable medium 206 also includes applications 208 or settings that are activated in response to the facial authentication system 202 authenticating the user 104. Exemplary applications 208 can include a password storage application, a banking application, a wallet application, a health application, or any application that provides user privacy.

[0027] The user device 102 can also include a network interface 210 for communicating data over wired, wireless, or optical networks. For example, the network interface 210 can communicate data over a local area network (LAN), a wireless local area network (WLAN), a personal area network (PAN), a wired local area network (WAN), an intranet, the Internet, a peer-to-peer network, a point-to-point network, a mesh network, and the like. The user device 102 can also include a display (not shown).

[0028] The facial authentication system 202 enables the user 104 to access the applications 208, settings, or other resources of the user device 102 using an image of the user’s 104 face. The facial authentication system 202 includes at least one camera system 212, at least one spoof detector 214, and at least one face identifier 216. Various implementations of the facial authentication system 202 can include a system on a chip (SoC), one or more integrated circuits (ICs), a processor with embedded processor instructions or configured to access processor instructions stored in memory, hardware with embedded firmware, a printed circuit board with various hardware components, or any combination thereof.

[0029] The facial authentication system 202 can be designed to operate under a variety of different environmental conditions. For example, the facial authentication system 202 can support facial authentication for distances between approximately 20 and 70 centimeters. These distances represent the distance between the user device 102 and the user 104. As another example, the facial authentication system 202 can support facial authentication for various tilt and pan angles of the user device 102 that provide an angular view between approximately -40 and 40 degrees. In addition, the facial authentication system 202 can be designed to make a decision regarding facial authentication within a predetermined amount of time, such as 100 to 200 milliseconds (ms). This time range can include the total amount of time taken to capture the image 106, process the image to determine whether a presentation attack 116 occurred, and process the image to determine whether the image 106 represents the face of the user 104.

[0030] The camera system 212 captures images 106 for facial authentication. The camera system 212 includes at least one camera, such as a red-green-blue (RGB) camera. The camera system 212 can also include one or more illuminators to provide illumination, particularly in dark environments. The illuminators can include RGB lights, such as LEDs. In some implementations, the camera system 212 can be used for other applications, such as for selfies, capturing pictures for the applications 208, scanning documents, reading barcodes, and the like.

[0031] The camera system 212 can have various customizable settings, including a capture rate and an exposure time. As an example, the camera system 212 can support a capture rate of 30 hertz (Hz) or more (e.g., 60 Hz, 90 Hz, 120 Hz, or 240 Hz). The camera system 212 can also support an exposure time of at least one millisecond (e.g., 1 ms, 10 ms, 100 ms, 0.1 second (sec), 1 sec, 5 sec, or 10 sec). By controlling the operational settings of the camera system 212, the facial authentication system 202 can facilitate pattern recognition for facial authentication spoof prevention. The facial authentication system 202 can appropriately balance the amount of light captured for pattern recognition with the exposure time. In some implementations, the facial authentication system 202 can consider other features of the camera system 212 that can enable additional light to be captured. These features can include a lens size of the camera system 212, available light sources, and techniques that together increase the capture rate and average frame.

[0032] The spoof detector 214 analyzes the images 106 captured using the camera system 212 and determines whether the images 106 depict a human face or a display 114 that presents a virtual face as part of a presentation attack 116. Specifically, the spoof detector 214 identifies whether a pattern associated with the operation of the display 114 is presented. Sometimes, the pattern can be identified based on a single image 106. Other times, the pattern can be identified based on across two or more images 106 or changes between two or more images 106. To increase confidence, the spoof detector 214 can analyze additional images (or sets of images) to confirm whether the pattern is presented.

[0033] If the pattern is presented, the spoof detector 214 determines that the user device 102 has suffered a presentation attack 116. If the pattern is not presented, the spoof detector 214 determines that the user device 102 has not suffered a presentation attack 116. The facial authentication system 202 can grant authorization in response to the spoof detector 214 not detecting a presentation attack 116.

[0034] Face recognizer 216 performs face recognition to verify that the face presented in image 106 corresponds to authorized user 104 of user device 102. Spoof detector 214 and face recognizer 216 can be implemented in software, programmable hardware, or some combination thereof. In some implementations, spoof detector 214 and / or face recognizer 216 are implemented using a machine learning module, such as a neural network. Regarding Figure 3 The operation of face authentication system 202 is further described.

[0035] Pattern recognition for face authentication spoofing

[0036] Figure 3 An exemplary face authentication system 202 that performs pattern recognition for face authentication spoofing is shown. In the depicted configuration, face authentication system 202 includes camera system 212, spoof detector 214, and face recognizer 216. Spoof detector 214 and face recognizer 216 are coupled to camera system 212. In some implementations, spoof detector 214 is coupled to face recognizer 216.

[0037] During operation, face authentication system 202 receives a request 302 to perform face authentication. In some cases, request 302 is provided by application 208 or other service that requires face authentication. In other cases, request 302 is provided by a sensor of user device 102 in response to detecting an interaction with user 104. For example, in response to detecting that user 104 is holding up user device 102, an inertial measurement unit (IMU) sensor sends request 302 to face authentication system 202.

[0038] In response to receiving request 302, face authentication system 202 initializes and activates camera system 212. In particular, face authentication system 202 sets an operating configuration 304 of camera system 212, such as a capture rate 306 (e.g., frame rate) and / or an exposure time 308. Operating configuration 304 is tailored so that face authentication system 202 is able to detect patterns associated with presentation attack 116 using display 114. In some implementations, operating configuration 304 is selected based on the likelihood that presentation attack 116 uses a particular type of display, such as an LED display on a smartphone. In other implementations, face authentication system 202 can cycle between different operating configurations 304 to improve the likelihood that face authentication system 202 detects presentation attack 116 performed by various different types of displays. At times, face authentication system 202 uses a first operating configuration 304 for a first time period to detect presentation attack 116 and switches to a second operating configuration 304 for a second time period to capture images for face recognition. Regarding Figure 4 to Figure 6 Exemplary patterns and operating configurations 304 are further described.

[0039] The camera system 212 captures the images 106-1 through 106-N using the operational configuration 304 specified by the face authentication system 202. The spoof detector 214 receives at least a portion of the images 106-1 through 106-N. By analyzing the images 106-1 through 106-N, the spoof detector 214 can detect a pattern associated with the presentation attack 116. The spoof detector 214 generates a pattern recognition indicator 310 that indicates whether the spoof detector 214 recognized the pattern within the images 106-1 through 106-N (e.g., indicates whether the spoof detector 214 detected the presentation attack 116). In some embodiments, the spoof detector 214 provides the pattern recognition indicator 310 to the face recognizer 216, as shown. In other implementations, the spoof detector 214 provides the pattern recognition indicator 310 to the application 208 or service that provided the request 302. Additionally or alternatively, the spoof detector 214 provides the pattern recognition indicator 310 to other logical components (not shown) of the face authentication system 202 that determine whether to provide authentication. Figure 3

[0040] The face recognizer 216 receives at least a portion of the images 106-1 through 106-N and the pattern recognition indicator 310. If the pattern recognition indicator 310 indicates that the presentation attack 116 was not detected, the face recognizer 216 performs face recognition to determine whether the face of the user 104 is presented within the images 106-1 through 106-N. The face recognizer 216 generates a report 312 that is provided to the application 208 or service that sent the request 302. If the face recognizer 216 recognizes the face of the user 104 and the pattern recognition indicator 310 indicates the absence of the presentation attack 116, the face recognizer 216 uses the report 312 to indicate a successful face authentication. Alternatively, if the face recognizer 216 does not recognize the face of the user 104 or the pattern recognition indicator 310 indicates the occurrence of the presentation attack 116, the face recognizer 216 uses the report 312 to indicate a failed face authentication. The spoof detector 214 can recognize various different patterns associated with the presentation attack 116, as further described with respect to Figure 4 to Figure 6

[0041] Figure 4 ​​An exemplary pulse width modulation pattern 400 that the spoof detector 214 can recognize is shown. Some displays 114 that include LED displays use pulse width modulation (PWM) to control the brightness level. Specifically, the displays 114 use a pulse width modulation signal to control the operating state of the LEDs within the display 114. Typically, the pulse width modulation signal rapidly pulses the LEDs on and off such that the human eye observes a display 114 that produces a steady dim light. The pulse width modulation signal can have two amplitude levels. The first amplitude level causes the LEDs to be in an on state that produces light. The second amplitude level causes the LEDs to be in an off state that does not produce light. In many cases, the pulse width modulation signal scans the display 114 from top to bottom, which creates dark regions across the display 114.

[0042] The pulse width modulation signal has a duty cycle and a frequency that determines the length of time that each LED produces light. An exemplary frequency of the pulse width modulation signal can be approximately 240 Hz or 360 Hz. In some displays 114, a pulse width modulation signal with a frequency of approximately 240 Hz can support a refresh rate that is approximately equal to 10, 30, 60, or 120 Hz. In addition, another pulse width modulation signal with a frequency of approximately 360 Hz can support a refresh rate that is approximately equal to 10 Hz, 30 Hz, 45 Hz, 60 Hz, 90 Hz, or 120 Hz. Exemplary duty cycles can be between 0% and 100%, including 20%, 50%, and 80%. Typically, decreasing the percentage of time that the LEDs are in the on state (e.g., decreasing the duty cycle of the pulse width modulation signal) darkens the display 114, and increasing the percentage of time that the LEDs are in the on state (e.g., increasing the duty cycle) brightens the display 114.

[0043] Using the pulse width modulation signal creates a pulse width modulation pattern 400 that is characterized by bright regions 402 interleaved with dark regions 404 (e.g., regions that are less bright than the bright regions 402). The width of the dark regions 404 varies based on the duty cycle of the pulse width modulation signal.

[0044] To detect the pulse width modulation pattern 400, the face authentication system 202 configures the camera system 212 to use a capture rate 306 that is greater than a first threshold for detecting the pulse width modulation pattern 400. The first threshold can be approximately 30 Hz or more (e.g., 90 Hz). As an example, the first threshold can be set to a frequency that is greater than or equal to an estimated refresh rate of the display 114. As described above, the estimated refresh rate of the display 114 can be associated with the frequency of the pulse width modulation signal.

[0045] Further, the facial authentication system 202 can configure the camera system 212 to use an exposure time 308 that is less than a second threshold used to detect the pulse width modulation pattern 400. The second threshold can be approximately equal to the inverse of an estimated frequency of the pulse width modulation signal. The estimated frequency can be a frequency that has a high likelihood of being used based on the type of display expected to be used in the presentation attack 116. For example, the exposure time 308 can be less than approximately 1 / 360 seconds or 1 / 240 seconds. Generally, the facial authentication system 202 can set the capture rate 306 and the exposure time 308 appropriately to enable a sufficient amount of light to be captured for detection of the pulse width modulation pattern 400.

[0046] During operation, the camera system 212 captures at least one of the images 106-1 to 106-3 of the display 114 that are presented as part of the presentation attack 116. The images 106-1 to 106-3 each include bright regions 402 and dark regions 404. In some cases, the bright regions 402 and the dark regions 404 occur at different locations within the images 106-1 to 106-3. By analyzing one or more of the images 106-1 to 106-3, the spoof detector 214 identifies the pulse width modulation pattern 400 and detects the presentation attack 116.

[0047] In an example, multiple ones of the images 106-1 to 106-3 are analyzed and changes between the images are identified, thereby allowing identification of the pulse width modulation pattern 400 and detection of the presentation attack 116. That is, identifying the pulse width modulation pattern 400 can include identifying the pattern within multiple images by comparing the different images and identifying changes between the images and / or by identifying different portions of the pattern in one or more of the images 106-1 to 106-3.

[0048] Figure 5 An exemplary black frame insertion pattern 500 that the spoof detector 214 is able to identify is shown. Some displays 114, including LCDs, LED displays, and some OLED displays, use black frame insertion (BFI) to reduce motion blur. This technique inserts a dark frame (e.g., a black frame or a dark color frame) between the refresh of the display 114 and the next frame. Generally, the dark frame is not as bright as the normal frame. The insertion of the dark frame effectively causes the human eye to observe a faster refresh rate, which adds additional smoothness and continuity to the motion depicted on the display 114.

[0049] Using black frame insertion, the black frame insertion pattern 500 is created, which is characterized by normal frames (e.g., normal frames 502-1 and 502-2) interleaved with dark frames (e.g., dark frames 504-1 and 504-2). Generally, the dark frames 504-1 and 504-2 are not as bright as the normal frames 502-1 and 502-2.

[0050] To detect the black frame insertion pattern 500, the facial authentication system 202 configures the camera system 212 to use a capture rate 306 that is greater than a threshold for detecting the black frame insertion pattern 500. The threshold can be twice the estimated refresh rate of the display 114. An example refresh rate can be approximately 30 Hz or more (e.g., 45 Hz, 60 Hz, 90 Hz, or 120 Hz). The estimated refresh rate can be the refresh rate that the type of display expected to be used in the presentation attack 116 has a high likelihood of being used.

[0051] During operation, the camera system 212 captures at least two of the images 106-1 to 106-4 of the display 114 that are presented as part of the presentation attack 116. The images 106-1 and 106-3 capture the normal frames 502-1 and 502-2 presented by the display 114. The images 106-2 and 106-4 capture the dark frames 504-1 and 504-2 presented by the display 114 due to black frame insertion. By analyzing two or more of the images 106-1 to 106-4, the spoof detector 214 identifies the black frame insertion pattern 500 and detects the presentation attack 116.

[0052] Figure 6 An example backlight scanning pattern 600 that the spoof detector 214 is able to identify is shown. Some displays 114 that include some LCDs use backlight scanning (backlight strobing or strobed backlight) to reduce motion blur. This technique turns off the backlight to darken different areas of the image in turn. Typically, the dark areas are not as bright as the rest of the frame.

[0053] The backlight scanning pattern 600 is created using backlight scanning, which is characterized by each frame having a dark area 602. The dark area 602 can vary in position between frames, as shown. Figure 6

[0054] To detect the backlight scanning pattern 600, the facial authentication system 202 configures the camera system 212 to use a capture rate 306 that is greater than a threshold for detecting the backlight scanning pattern 600. The threshold can be twice the estimated backlight scanning rate of the display 114. The backlight scanning rate can be approximately 30 Hz or more (e.g., 45 Hz, 60 Hz, 90 Hz, or 120 Hz). The estimated backlight scanning rate can be the rate that the type of display expected to be used in the presentation attack 116 has a high likelihood of being used.

[0055] ​During operation, the camera system 212 captures at least one of the images 106-1 through 106-4 of the display 114 that are presented as part of the presentation attack 116. The images 106-1 through 106-4 each include a dark region 602 that encompasses a portion of the images 106-1 through 106-4. When the images 106-1 through 106-4 are organized sequentially, the dark region 602 appears to move down across the images 106-1 through 106-4. For example, the dark region 602 is at the top of the image 106-1, while the image 106-4 has the dark region 602 near the middle-lower position. By analyzing one or more of the images 106-1 through 106-4, the spoof detector 214 identifies the backlight scanning pattern 600 and detects the presentation attack 116.

[0056] In an example, multiple ones of the images 106-1 through 106-4 are analyzed and changes between the images are identified, allowing the backlight scanning pattern 600 to be identified and the presentation attack 116 to be detected. That is, identifying the backlight scanning pattern 600 can include identifying the pattern within multiple images by comparing different images and identifying changes and / or by identifying different portions of the pattern in one or more of the images 106-1 through 106-4.

[0057] Exemplary method

[0058] Figure 7 An exemplary method 700 for performing operations for pattern recognition for face authentication anti-spoofing is depicted. The method 700 is shown as a collection of operations (or actions) performed, but is not necessarily limited to the order or combination of these operations shown herein. Further, any one or more of these operations can be repeated, combined, reorganized, or linked to provide a wide array of additional and / or alternative methods. In the sections of the following discussion, reference can be made to the environment 100 and the entities detailed in Figure 1 or Figure 2 or Figure 3 which are merely examples. The techniques are not limited to being performed by one entity or multiple entities operating on one device.

[0059] At 702, a capture rate of a camera system of a face authentication system is set to be greater than a threshold for detecting a pattern associated with operation of a display. For example, the face authentication system 202 sets the capture rate 306 of the camera system 212 to be greater than a threshold for detecting a pattern associated with operation of the display 114. The pattern can be the pulse width modulation pattern 400 of Figure 4 FIG. 4, which is associated with brightness of the display 114. Further, the pattern can be the black frame insertion pattern 500 or the backlight scanning pattern 600 associated with motion blur reduction.

[0060] At 704, at least one image of the object is captured using a capture rate. For example, the camera system 212 of the facial authentication system 202 captures one or more of the images 106-1 through 106-N using the capture rate 306, as shown in Figure 3 FIG. 6. The images 106-1 through 106-N depict the object. In some cases, the object is the display 114 used as part of the presentation attack 116, as shown in Figure 1 FIG. 6.

[0061] At 706, a pattern is identified within the at least one image. For example, the spoof detector 214 of the facial authentication system 202 identifies a pattern within at least one of the images 106-1 through 106-N. Specifically, the spoof detector 214 can identify the pulse width modulation pattern 400 of Figure 4 FIG. 4 or the backlight scanning pattern 600 of Figure 6 FIG. 5 within one or more of the images 106-1 through 106-N. Alternatively, the spoof detector 214 can identify the pattern as the black frame insertion pattern 500 of Figure 5 FIG. 5 based on a change across two or more of the images 106-1 through 106-N or a change between two or more of the images.

[0062] At 708, in response to identifying the pattern, the object is determined to include a display. For example, in response to identifying the pattern, the spoof detector 214 determines that the object includes at least a portion of the display 114. In this case, the spoof detector 214 generates the pattern identification indicator 310 to indicate that the presentation attack 116 was detected, as shown in Figure 3 FIG. 6.

[0063] At 710, in response to determining that the object includes a display, facial authentication is prevented. For example, in response to determining that the object includes the display 114, the spoof detector 214 prevents facial authentication. The spoof detector 214 can prevent facial authentication by causing the face recognizer 216 to not perform face recognition or by causing the facial authentication system 202 to generate the report 312 indicating that the facial authentication failed.

[0064] In some implementations, the method 700 represents a portion of a complete facial authentication process. For example, the facial authentication system 202 can also perform face recognition using the face recognizer 216 prior to authenticating the user 104. In some cases, the steps 702 through 710 can occur prior to capturing other images for face recognition. For face recognition, the capture rate 306 (and other settings) of the camera system 212 can be set in a different manner than described at 702.

[0065] Example Computing System

[0066] Figure 8Various components of an example computing system 800 are shown, which can be implemented as any of the types of client, server, and / or computing devices described earlier Figure 2 to implement a facial authentication system 202 that performs pattern recognition for facial authentication anti-spoofing.

[0067] Computing system 800 includes a communication device 802 enabling wired and / or wireless communication of device data 804 (e.g., received data, data that is being received, data scheduled for broadcast, or data packets of the data). The device data 804 or other device content can include configuration settings of the device, media content stored on the device, and / or information associated with a user of the device 104. Media content stored on computing system 800 can include any type of audio, video, and / or image data. Computing system 800 includes one or more data inputs 806 via which any type of data, media content, and / or inputs can be received, such as human utterances, user-selectable inputs (explicit or implicit), messages, music, television media content, recorded video content, and any other type of audio, video, and / or image data received from any content and / or data source.

[0068] Computing system 800 also includes a communication interface 808 that can be implemented as any one or more of a serial and / or parallel interface, a wireless interface, any type of network interface, any type of modem, and the like, as well as any other type of communication interface. The communication interface 808 provides a connection and / or communication links between computing system 800 and a communication network by which other electronic, computing, and communication devices communicate data with computing system 800.

[0069] Computing system 800 includes one or more processors 810 (e.g., any of microprocessors, controllers, and the like) which process various computer- executable instructions to control the operation of computing system 800 and to implement techniques for gesture recognition in the presence of saturation or in which the techniques for gesture recognition in the presence of saturation can be implemented. Alternatively or additionally, computing system 800 can be implemented with any one or combination of hardware, firmware, or fixed logic circuitry that is

[0070] The computing system 800 also includes computer-readable media 814, such as one or more memory devices that implement a persistent and / or non-transitory data storage (i.e., in contrast to mere signal transmission), examples of which include random access memory (RAM), non-volatile memory (e.g., any one or more of a read-only memory (ROM), flash memory, EPROM, EEPROM, or the like), and a disk storage device. A disk storage device can be implemented as any type of magnetic or optical storage device, such as a hard disk drive, a recordable and / or rewritable compact disc (CD), any type of a digital versatile disc (DVD), and the like. The computing system 800 can also include a camera system 212 and a mass storage media device (storage media) 816.

[0071] The computer-readable media 814 provides data storage mechanisms to store the device data 804, as well as various device applications 818 and any other types of information and / or data related to operational aspects of the computing system 800. For example, an operating system 820 can be maintained as a computer application with the computer-readable media 814 and executed on the processors 810. The device applications 818 can include a device manager, such as any form of a control application, software application, signal-processing and control module, code that is native to a particular device, a hardware abstraction layer for a particular device, and so on.

[0072] The device applications 818 also include any system components, engines, or managers to implement pattern recognition for face authentication anti-spoofing. In the present example, the device applications 818 include a spoof detector 214 and a face recognizer 216 of the face authentication system 202. Figure 2 ) of the face authentication system 202.

[0073] CONCLUSION

[0074] Although the technology using pattern recognition for face authentication anti-spoofing and the apparatus including pattern recognition for face authentication anti-spoofing have been described in language specific to structural features and / or methods, it is to be understood that the subject of the appended claims is not necessarily limited to the specific features or methods described. Rather, the specific features and methods are disclosed as illustrative examples of pattern recognition for face authentication anti-spoofing.

[0075] Some examples are described below.

[0076] Example 1 : A method performed by a face authentication system, the method comprising:

[0077] setting a capture rate of a camera system of the face authentication system to be greater than a threshold for detecting a pattern associated with operation of a display;

[0078] capturing at least one image of a subject using the capture rate;

[0079] identifying the pattern within the at least one image;

[0080] in response to identifying the pattern, determining that the object includes the display; and

[0081] in response to determining that the object includes the display, preventing face authentication.

[0082] Example 2: The method of example 1, further comprising:

[0083] capturing at least one second image of a second object using the capture rate;

[0084] determining that the pattern is not present within the at least one second image;

[0085] identifying the second object as a face of a user; and

[0086] in response to determining that the pattern is not present and identifying the second object, authenticating the user.

[0087] Example 3: The method of example 1 or 2, wherein the pattern includes bright regions interleaved with dark regions within the at least one image.

[0088] Example 4: The method of example 3, wherein the pattern includes a pulse width modulation pattern associated with a brightness of the display.

[0089] Example 5: The method of example 3 or 4, wherein the threshold for detecting the pulse width modulation pattern is approximately 30 hertz.

[0090] Example 6: The method of any preceding example, further comprising controlling an exposure time of the camera system to be less than a second threshold for detecting the pattern.

[0091] Example 7: The method of example 6, wherein the second threshold is approximately equal to an inverse of an estimated frequency of a pulse width modulation signal associated with the display.

[0092] Example 8: The method of example 1 or 2, wherein the pattern includes dark regions within the at least one image.

[0093] Example 9: The method of example 8, wherein:

[0094] the at least one image includes a plurality of images; and

[0095] the dark regions comprise a total area of each image within a subset of the plurality of images.

[0096] Example 10: The method of example 9, wherein the pattern comprises a black frame insertion pattern associated with motion blur reduction.

[0097] Example 11 : The method of example 10, wherein the threshold for detecting the black frame insertion pattern is greater than twice a refresh rate of the display.

[0098] Example 12: The method of example 8, wherein the dark region comprises a portion of the at least one image.

[0099] Example 13: The method of example 12, wherein the pattern comprises a backlight scanning pattern associated with motion blur reduction.

[0100] Example 14: The method of example 13, wherein the threshold for detecting the backlight scanning pattern is greater than twice a backlight scanning rate of the display.

[0101] Example 15: The method of any preceding example, wherein:

[0102] the at least one image comprises a plurality of images; and

[0103] identifying the pattern within the at least one image comprises identifying the pattern within the plurality of images.

[0104] Example 16: The method of example 15, wherein identifying the pattern within the plurality of images comprises identifying changes between images of the plurality of images.

[0105] Example 17: The method of any preceding example, wherein the display comprises:

[0106] a light emitting diode display;

[0107] a liquid crystal display; or

[0108] another liquid crystal display with backlight scanning.

[0109] Example 18: An apparatus comprising a facial authentication system configured to perform the method of any of examples 1 to 17.

[0110] Example 19: A computer-readable storage medium comprising instructions that, in response to execution by a processor, cause a facial authentication system to perform the method of any of examples 1 to 17.

Claims

1. A method performed by a facial authentication system, the method comprising: setting a capture rate of a camera system of the facial authentication system to be greater than a threshold for detecting a pattern associated with operation of a display; capturing at least one image of an object using the capture rate; identifying the pattern within the at least one image; in response to identifying the pattern, determining that the object comprises the display; and in response to determining that the object comprises the display, preventing facial authentication.

2. The method of claim 1, further comprising: capturing at least one second image of a second object using the capture rate; determining that the pattern is not presented within the at least one second image; identifying the second object as a face of a user; and in response to determining that the pattern is not presented and identifying the second object, authenticating the user. the pattern comprises bright regions interleaved with dark regions within the at least one image. the pattern comprises a pulse width modulation pattern associated with brightness of the display.

3. The method of claim 1, wherein, the threshold for detecting the pulse width modulation pattern is 30 hertz.

4. The method of claim 3, wherein, controlling an exposure time of the camera system to be less than a second threshold for detecting the pattern.

5. The method of claim 4, wherein, the second threshold is equal to an inverse of an estimated frequency of a pulse width modulation signal associated with the display.

6. The method of claim 1, further comprising: the pattern comprises dark regions within the at least one image.

7. The method of claim 6, wherein, 9. The method of claim 8, wherein:

8. The method of claim 1, wherein, the at least one image comprises a plurality of images; and the dark regions comprise a total area of each image within a subset of the plurality of images. the pattern comprises a black frame insertion pattern associated with motion blur reduction. the threshold for detecting the black frame insertion pattern is greater than twice a refresh rate of the display.

10. The method of claim 9, wherein, the dark regions comprise a portion of the at least one image.

11. The method of claim 10, wherein, the pattern comprises a backlight scanning pattern associated with motion blur reduction.

12. The method of claim 8, wherein, the threshold for detecting the backlight scanning pattern is greater than twice a backlight scanning rate of the display.

13. The method of claim 12, wherein, 15. The method of any of claims 1-14, wherein:

14. The method of claim 13, wherein, the at least one image comprises a plurality of images; and identifying the pattern within the at least one image comprises identifying the pattern within the plurality of images. identifying the pattern within the plurality of images comprises identifying changes between images of the plurality of images. the display comprises:

16. The method of claim 15, wherein, a light emitting diode display; 17. The method of any one of claims 1-14, wherein, a liquid crystal display; or another liquid crystal display with backlight scanning.

18. An apparatus comprising a facial authentication system configured to perform the method of any of claims 1-17.

19. A computer-readable storage medium comprising instructions that, in response to execution by a processor, cause a facial authentication system to perform the method of any of claims 1-17. ​ ​

Images