Living body detection method based on near infrared and remote photoplethysmography

Abstract

The invention relates to a living body detection method based on near infrared and remote photoplethysmography, and belongs to the technical field of calculation, reckoning and counting. The method comprises the following steps: positioning a face region of a to-be-detected object for face recognition; obtaining a near-infrared light image of the face of the object passing the face recognition, and obtaining an infrared light image as a score of a reproduced self-screen medium by detecting the near-infrared light image; and acquiring a face image sequence of the object detected through screencopying, extracting vital sign signals from the to-be-detected image sequence by using remote photoplethysmography, and identifying whether the to-be-detected object is a living body face. According to the method, the robustness of living body detection is improved, the method has stronger classification and learning capabilities, representation attacks such as three-dimensional masks, videos andphotograph copying can be better coped with, and the distinguishing result accuracy is higher. The improved remote photoplethysmography of superposing the shortest period signal to extract the vital sign signal does not need priori knowledge, and can adapt to the requirements of an actual use scene.

Description

technical field [0001] The invention discloses a living body detection method based on near-infrared and long-distance photoplethysmography, relates to face recognition technology, and belongs to the technical field of calculation, calculation and counting. Background technique [0002] With the development of neural network, face recognition technology based on feature learning is becoming more and more mature. Today, face recognition technology has been widely used in access control systems, electronic payments, stations and many other scenarios that require identity authentication. However, traditional face recognition algorithms only focus on identifying the identity information of the object to which the face belongs, and pay little attention to whether the face to be detected comes from a real object. When faced with attack methods such as photos of legitimate users, pre-recorded videos, and 3D masks, traditional face recognition systems cannot determine whether the o...

AI Technical Summary

Problems solved by technology

However, this method relies on a large number of data sets, a complex network structure, and a large number of computing resources, which is difficult to deploy in mobile terminals and difficult to adapt to the current trend

[0007] At present, there is still a scheme that uses remote photoplethysmography to realize face liveness detection. This scheme has the defect of misidentifying the vital sign signal when facing an image sequence attack such as remake video, which will cause the problem of misidentification.

[0008] To sum up, the existing human face liveness detection methods have their own defects, and there are problems that lead to poor liveness detection results, which cannot meet the needs of actual use.

Images