A static three-dimensional face in-vivo detection method based on deep learning

Abstract

The invention discloses a static three-dimensional face in-vivo detection method based on deep learning, and the method comprises the steps: shooting a color image through a color camera, and shootinga depth image through a depth camera; when a face is detected in the color image, obtaining a feature vector corresponding to the face through a first convolutional neural network and a second convolutional neural network according to the color image and the depth image; and judging whether the face is a living body or not through a pre-trained living body detection classifier according to the feature vector corresponding to the face. Only two cameras are adopted, the characteristics of color images and depth images are combined, the technologies of deep learning, machine learning and the like are combined, the speed, the passing rate and the anti-counterfeiting rate of human face living body detection are greatly increased, the cost is low, and the precision is high.

Description

technical field [0001] The invention belongs to the technical field of face recognition, and in particular relates to a static three-dimensional human face detection method based on deep learning. Background technique [0002] With the continuous development of face recognition technology, many products have begun to use face recognition technology to verify user identities, such as bank ATM machines, unmanned stores and even home door locks. However, general face recognition technology cannot effectively detect whether the user is alive or not. Therefore, malicious people can pretend to be legitimate users by printing other people's photos or using mobile phones to shoot other people's videos, and deceive the face recognition system to achieve their malicious purposes. Therefore, face detection technology came into being. [0003] Currently, a face liveness detection technology is provided in the related art, which uses images captured by two cameras to obtain 3D face feat...

AI Technical Summary

Problems solved by technology

[0004] In the above-mentioned related technologies, the data of two cameras is used to obtain 3D face feature points, which takes a long time and cannot achieve the purpose of real-time detection. At the same time, it relies on the accuracy of the human eye detection algorithm, which cannot guarantee efficiency and accuracy.

Moreover, three cameras are required, and the angle of view and image alignment of the three cameras must be considered at the same time, which requires a large amount of calculation and high cost.

Images