A novel cost-effective vivo facial detection technique based on snapshot spectral imaging sensor

Abstract

In the field of face anti-counterfeiting, there are differences between the reflection spectrum of real faces and simulated faces, which can help us overcome the shortcomings of traditional RGB cameras that are difficult to identify the authenticity of faces. In our work, we designed a face anti-spoofing imaging system based on the snapshot spectral imaging chip, which can be used in face anti-spoofing imaging through the analysis of spectral imaging data. Experiments show that our sensor could reconstruct the spectrum of the face region, establish the spectral databases, and achieve face authenticity recognition under active light source based on deep convolutional neural network, with a face recognition accuracy of 95%.