A novel contrast enhancement forensics based on convolutional neural networks

作者：

Highlights：

•

• A novel CNN-based contrast enhancement (CE) forensic method is proposed.

• Gray level co-occurrence matrix (GLCM) is fed into the CNN.

• The CNN trained with GLCM is superior to the one trained using the image input.

• The proposed method is robust to state-of-the-art anti-forensic attacks.

• The proposed method outperforms the conventional CE forensic approaches.

摘要

•A novel CNN-based contrast enhancement (CE) forensic method is proposed.•Gray level co-occurrence matrix (GLCM) is fed into the CNN.•The CNN trained with GLCM is superior to the one trained using the image input.•The proposed method is robust to state-of-the-art anti-forensic attacks.•The proposed method outperforms the conventional CE forensic approaches.

论文关键词：Digital image forensics,Contrast enhancement,Convolutional neural networks,Deep learning,Gray level co-occurrence matrix

论文评审过程：Received 27 September 2017, Revised 11 January 2018, Accepted 1 February 2018, Available online 12 February 2018, Version of Record 16 March 2018.

论文官网地址：https://doi.org/10.1016/j.image.2018.02.001