RGB (Red Green Blue) and IR (Infrared) binocular camera-based living body detecting method and device

Abstract

The invention relates to an RGB (Red Green Blue) and IR (Infrared) binocular camera-based living body detecting method and device. The method comprises the steps of obtaining two sets of video streamsthrough an RGB camera and an IR camera respectively; carrying out face detection and living body judgement on video frames in the two sets of video streams; and when both the two sets of video framesare judged as living bodies, regarding the face in the current video frame to be a living human face. The method disclosed by the invention specifically comprises the steps of collecting face videosby adopting the two cameras and carrying out face detection to respectively obtain human faces under RGB and IR; aiming at an RGB colorful face image, extracting LBP (Local Binary Pattern) features byutilizing a traditional image processing algorithm and judging the face is a living human face or not through SVM (Support Vector Machine) classification; meanwhile, aiming at an IR face image, directly entering a trained CNN (convolutional neural network) to carry out classification and judge whether the face is a living human face or not; and if both the faces are judged as living human faces,eventually judging the face is a living face. The method disclosed by the invention has the beneficial effects of high robustness, low cost and convenience for large-scale use.

Description

technical field [0001] The invention relates to the technical field of face recognition, in particular to a living body detection method and device based on RGB and IR binocular cameras. Background technique [0002] As a biometric technology, face recognition has been widely used in security, new retail, mobile payment and other fields due to the rapid development of artificial intelligence in recent years. Compared with other biometrics, such as fingerprint recognition and iris recognition, it is more Direct, friendly and convenient. [0003] In order to effectively detect the authenticity and effectiveness of face identity, it mainly includes the following methods: [0004] 1. Analysis based on texture, frequency, optical flow field and other characteristics [0005] Using traditional image processing algorithms, LBP or optical flow fields are used to analyze the features of live and non-living face images, and then SVM is used to classify and judge. [0006] 2. Analys...

AI Technical Summary

Problems solved by technology

[0012] However, using traditional image processing to analyze the image texture, optical flow field, etc., it is possible to use high-definition images, or retinal screen images and videos, and there is a high probability of breaches.

Although 3D cameras can effectively prevent 2D plane tools, the cost is too high to be commercially used in different fields on a large scale, such as access control systems.

Images