According to the difference between real images and false photos, texture and statistical feature are extracted in different ways, this project trains SVM classifers with features below:
DoG (Difference of Gaussian);
LBP (Local Binary Pattern);
HSV histograms;
HOOF (Histograms of Optical Flows).
The dataset used are:
NUAA;
CASIA_FASD;
REPLAY-ATTACK.
The features and datasets are combined with each other in different ways by setting control groups. The details can be found here.
Face Liveness Detection is published in 3 languages.
C++ Version.
You can train your SVM classifier and deploy it on the server for work.
MATLAB Version.
You can train different classifiers by setting control groups and analyze the correct rate.
You can learn how to set control groups in this page.
Citation:
[1] HuaCheng Liu. The Gordian Technique research of Face Liveness Detection[D]. NingBo University 2014.
[2] REPLAY-ATTACK Database.
@INPROCEEDINGS{Chingovska_BIOSIG-2012,
author = {Chingovska, Ivana and Anjos, Andr{\'{e}} and Marcel, S{\'{e}}bastien},
keywords = {biometric, Counter-Measures, Local Binary Patterns, Spoofing Attacks},
month = september,
title = {On the Effectiveness of Local Binary Patterns in Face Anti-spoofing},
journal = {IEEE BIOSIG 2012},
year = {2012}
}
[3] CASIA-FASD Database.
@INPROCESSINGS{zhang2012face,
title = {A face antispoofing database with diverse attacks},
author = {Zhang, Zhiwei and Yan, Junjie and Liu, Sifei and Lei, Zhen and Yi, Dong and Li, Stan Z},
booktitle = {Biometrics (ICB), 2012 5th IAPR international conference on},
pages = {26--31},
year = {2012},
organization = {IEEE}
}
[4] NUAA Database.
[5] HOOF Toolbox.
R. Chaudhry, A. Ravichandran, G. Hager and R. Vidal.
Histograms of Oriented Optical Flow and Binet-Cauchy Kernels on Nonlinear Dynamical Systems for the Recognition of Human Actions.
CVPR, 2009.
[6] LibSVM Toolovbox.
References:
License:
MIT License.
Author:
Hai-Liang Zhao (hliangzhao97@gmail.com);