Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (6): 170-176.DOI: 10.3778/j.issn.1002-8331.2108-0046

• Pattern Recognition and Artificial Intelligence • Previous Articles     Next Articles

Face Liveness Detection Based on Fusional Optical Flow and Texture Features

WANG Hongfei, CHENG Xin, ZHAO Xiangmo, ZHOU Jingmei   

  1. 1.School of Information Engineering, Chang’an University, Xi’an 710064, China
    2.School of Electronic and Control Engineering, Chang’an University, Xi’an 710064, China
  • Online:2022-03-15 Published:2022-03-15

光流与纹理特征融合的人脸活体检测算法

王宏飞,程鑫,赵祥模,周经美   

  1. 1.长安大学 信息工程学院,西安 710064
    2.长安大学 电子与控制工程学院,西安 710064

Abstract: Aiming at video replaying, the common face recognition system fraud method, this paper utilizes the semantic information of face attack images to propose a face spoofing detection algorithm based on optical flow and texture features. The optical flow field map of the face area is generated by the optical flow method and face detection method from two consecutive frames of the captured face video. Then, the original RGB face area image and optical flow field map are input to a 2-channels convolutional neural network to extract and fuse the features of the face. Finally, based on the optical flow and texture features, it realizes classification of real and fake faces. In addition, in order to better generate the optical flow field map containing liveness information, a motion amplification algorithm is applied to enhance the 0.04-0.4 Hz signal in the video frame by 20 times. This paper uses the Replay Attack spoofing dataset from IDIAP consisting of 1,300 videos for model training, verification and testing. Experiments show that the proposed algorithm performs well on the Replay Attack data set and achieves half total error rate of 1.04%.

Key words: computer vision, neural network, spoofing detection, optical flow

摘要: 针对照片与视频重放这一常见人脸识别欺诈手段,利用人脸攻击图像的语义信息提出一种基于光流与纹理特征融合的人脸活体检测算法:采集连续两帧待检测人脸图像,通过光流法及人脸检测方法生成人脸区域光流场变化图,将其与原始RGB图像输入至2通道卷积神经网络提取并融合得到人脸动-静态特征,基于融合特征实现真实人脸与欺诈人脸分类。此外,为了更好地捕捉人脸区域的光流场变化,应用影像动作放大技术将待检测视频帧中0.04~0.4?Hz信号放大20倍。使用IDIAP的Replay Attack人脸活体检测数据集中1?300段视频数据进行模型训练、验证和测试。实验表明,提出的人脸活体检测算法在Replay Attack数据集测试表现良好并取得了1.04%半错误率,能够有效识别照片、视频攻击。

关键词: 计算机视觉, 神经网络, 活体检测, 光流法