Face anti-spoofing detection based on multi-scale image quality assessment
作者：
Highlights：
• Pioneering usage of multiscale image quality features for face liveness detection.
• Three unique maps of the local face trait, local face trait product, and top gradient similarity derived from face images.
• Two statistical models along with a new top gradient similarity deviation model for face feature extraction.
• Six benchmark datasets based on intra-dataset and cross-dataset protocols for performance evaluation.
• Fair comparison with a number of state-of-the-art methods demonstrates the superiority.
摘要
•Pioneering usage of multiscale image quality features for face liveness detection.•Three unique maps of the local face trait, local face trait product, and top gradient similarity derived from face images.•Two statistical models along with a new top gradient similarity deviation model for face feature extraction.•Six benchmark datasets based on intra-dataset and cross-dataset protocols for performance evaluation.•Fair comparison with a number of state-of-the-art methods demonstrates the superiority.
论文关键词：Face recognition,Image quality assessment,Face presentation attack detection,Local face trait,Top gradient similarity
论文评审过程：Received 19 November 2020, Revised 9 December 2021, Accepted 1 March 2022, Available online 12 March 2022, Version of Record 21 March 2022.
论文官网地址：https://doi.org/10.1016/j.imavis.2022.104428
