一种基于空域和频域多特征融合的人脸活体检测算法

Face spoofing detection via integrating multiple features in spatial and frequency domains

为提高人脸识别技术的安全性,提出一种融合多种图像空域特征和频域特征的人脸活体检测算法。该算法在图像空域提取人脸的色相矩特征和模糊度特征,在频域提取人脸的傅里叶谱能量特征和能量占比特征,再将空域和频域特征进行级联融合并做归一化处理作为描述人脸图像的全局特征向量,用于训练SVM分类器并区分活体人脸和伪造人脸。在两个公开的数据库NUAA和CASIA上的实验结果表明,该算法性能优于其它基于单一特征或单一域特征的人脸活体检测算法。

To improve the security of face recognition system,a face spoofing detection algorithm is proposed which integrates multiple image features in spatial and frequency domains.Firstly,the chromatic moment features and blurriness features are extracted from the face image in the spatial domain,and the energy features and the energy proportion features of image Fourier spectrum are extracted in the frequency domain.Then,these spatial and frequency domain features are cascaded and normalized as the holistic feature vector to represent a face image.Finally,the face feature vectors are used for training the SVM classifier and to distinguish genuine faces from spoofing faces.Experimental results on two pubic databases NUAA and CASIA indicate that the proposed algorithm is superior to other algorithms based on single feature or single domain features in face spoofing detection.