CSC-Net: Cross-Color Spatial Co-Occurrence Matrix Network for Detecting Synthesized Fake Images
- 9 May 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 16 (1) , 369-379
Abstract
Recently, the Generative Adversarial Networks (GAN) generated images have been spread over the social networks, which brings the new challenge in the the community of media forensics. Although some reliable forensic tools have advanced the study of detecting GAN generated images, while the detection accuracy cannot be guaranteed when facing the malicious post-processing attacks, especially in the practical social network scenario. Thus, in this context, we propose a novel well-designed deep neural network equipped with handed-crafted features for dealing with this problem. In particular, relying on the Cross-color Spatial Co-occurrence Matrix (CSCM), the discriminative features are extracted after carefully analyzing and selecting the most effective color channels. Next, the fused features are fed into the deep neutral network for training a high-efficient forensic detector. Extensive experimental results empirically verify that in most detection scenarios, our proposed detector performs superiorly to the prior arts, especially in the case of post-processing attacks. Moreover, we also highlight the relevance of the proposed detector over the realistic social network platforms, and its generalization capability in three different scenarios.Keywords
Funding Information
- Zhejiang Provincial Natural Science Foundation of China (LZ23F020006)
- Fundamental Research Funds for the Provincial Universities of Zhejiang (GK219909299001-007)
- Open Projects Program of National Laboratory of Pattern Recognition
- Open Foundation of Henan Key Laboratory of Cyberspace Situation Awareness (HNTS2022016)
- National Key Research and Development Program of China (2022YFB3102900)
- National Natural Science Foundation of China (U1804263, 62172435, 62271180)
- Zhongyuan Science and Technology Innovation Leading Talent Project of China (214200510019)
This publication has 42 references indexed in Scilit:
- Distinguishing between natural and recolored images via lateral chromatic aberrationJournal of Visual Communication and Image Representation, 2021
- Identification of deep network generated images using disparities in color componentsSignal Processing, 2020
- On the Generalization of GAN Image Forensics2019
- Identifying Individual Camera Device From RAW ImagesIEEE Access, 2018
- Source camera identification via low dimensional PRNU featuresMultimedia Tools and Applications, 2018
- Robust Multi-Classifier for Camera Model Identification Based on Convolution Neural NetworkIEEE Access, 2018
- Individual camera device identification from JPEG imagesSignal Processing: Image Communication, 2017
- Exposing image resampling forgery by using linear parametric modelMultimedia Tools and Applications, 2017
- Passive forensics in image and video using noise features: A reviewDigital Investigation, 2016
- ImageNet: A large-scale hierarchical image database2009