Systems and methods for spoof detection and liveness analysis

Abstract

Spoof-detection and liveness analysis is performed using a software-based solution on a user device, such as a smartphone having a camera, audio output component (e.g., earpiece), and audio input component (e.g., microphone). One or more audio signals are emitted from the audio output component of the user device, reflect off a target, and are received back at the audio input component of the device. Based on the reflections, a determination is made as to whether the target is comprised of a three-dimensional face-like structure and / or face-like tissue. Using at least this determination, a finding is made as to whether the target is likely to be spoofed, rather than a legitimate, live person.

Description

[0001] Cross References to Related Applications [0002] This application claims priority and benefit to U.S. Provisional Patent Application No. 62 / 180,481, filed June 16, 2015, and entitled "Liveness Analysis Using Vitals Detection," which states The US Patent Application is hereby incorporated by reference in its entirety. technical field [0003] The present invention relates generally to image, acoustic signal, and vibration signal analysis, and in particular, the present invention relates to image and signal processing techniques for detecting whether a subject depicted in an image is alive. Background technique [0004] Often it may be desirable to restrict access to attributes or resources to specific individuals. Biometric systems can be used to authenticate an individual's identity to grant or deny access to resources. For example, iris scanners may be used by biometric security systems to identify individuals based on unique structures in their irises. However,...

AI Technical Summary

Problems solved by technology

However, these measures can only detect static (e.g., image-based) impersonation attacks, and require some time window of observations to be observed at a high frame rate sufficient to be able to resolve the aforementioned motion vectors into their required velocity and frequency distributions (if any)

Said measures can also erroneously reject live bodies that remain completely still during the scan or erroneously accept static reproductions with additional motion due to, for example, bending and shaking the fake photo in certain ways

[0006] Existing anti-counterfeiting measures of the second category assume that the photo or video reproduction of the biometric sample is of poor quality and that image texture analysis methods can thus give the identity of the counterfeiter

Another category of anti-counterfeiting measures related to the second category (which is based on reference or non-reference image quality metrics) suffers from the same disadvantages

Images