Comparative analysis of feature extraction and fusion for blind authentication of digital images using chroma channels

作者：

Highlights：

• Evaluate and benchmark different models of feature extraction to detect digitally-altered images.

• Combine various features utilizing single and multi-scale representations to enhance recognition accuracy.

• Intensively compare and statistically analyze different models integrating the three color channels.

• Assess the impact on the performance of two feature reduction methods using PCA and LLP.

摘要

•Evaluate and benchmark different models of feature extraction to detect digitally-altered images.•Combine various features utilizing single and multi-scale representations to enhance recognition accuracy.•Intensively compare and statistically analyze different models integrating the three color channels.•Assess the impact on the performance of two feature reduction methods using PCA and LLP.

论文关键词：Image forensics,Feature extraction,Dimensionality reduction,Fusion methods,Multimodal,Machine learning,Support vector machines

论文评审过程：Received 28 May 2020, Revised 4 January 2021, Accepted 6 April 2021, Available online 16 April 2021, Version of Record 21 April 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116271