Detection of spoofing attacks for ear biometrics through image quality assessment and deep learning
作者：
Highlights：
• A novel ear anti-spoofing method using the fusion of two systems is proposed.
• Image Quality Measures (IQM) are used and combined with score-level fusion.
• Deep learning based CNN architecture is employed for ear anti-spoofing.
• Decision-level fusion of IQM and CNN achieves zero error rates on five datasets.
• Results are comparable with biometrics-based state-of-the-art anti-spoofing systems.
摘要
•A novel ear anti-spoofing method using the fusion of two systems is proposed.•Image Quality Measures (IQM) are used and combined with score-level fusion.•Deep learning based CNN architecture is employed for ear anti-spoofing.•Decision-level fusion of IQM and CNN achieves zero error rates on five datasets.•Results are comparable with biometrics-based state-of-the-art anti-spoofing systems.
论文关键词：Ear biometrics,Spoof detection,Printed photo attack,Image quality measure,Deep learning
论文评审过程：Received 7 August 2019, Revised 10 January 2021, Accepted 10 January 2021, Available online 15 January 2021, Version of Record 11 February 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.114600
