CNN spatiotemporal features and fusion for surveillance video forgery detection

作者：

Highlights：

• A novel video forgery detection system using 2D-CNN and SSIM is proposed.

• A new, efficient feature extraction algorithm based on STP images is proposed.

• SSIM fusion is proposed to produce the feature of a whole video.

• The proposed system is more efficient and robust than previous forgery detection systems.

• The proposed system is highly usable in a real-time video authentication system.

摘要

•A novel video forgery detection system using 2D-CNN and SSIM is proposed.•A new, efficient feature extraction algorithm based on STP images is proposed.•SSIM fusion is proposed to produce the feature of a whole video.•The proposed system is more efficient and robust than previous forgery detection systems.•The proposed system is highly usable in a real-time video authentication system.

论文关键词：Passive forensics,Convolution neural network,SSIM,Spatiotemporal features,Inter-frame forgeries

论文评审过程：Received 4 July 2019, Revised 30 October 2020, Accepted 4 November 2020, Available online 8 November 2020, Version of Record 16 November 2020.

论文官网地址：https://doi.org/10.1016/j.image.2020.116066