TY - GEN
T1 - DETECTION OF FAKE IMAGES VIA THE ENSEMBLE OF DEEP REPRESENTATIONS FROM MULTI COLOR SPACES
AU - He, Peisong
AU - Li, Haoliang
AU - Wang, Hongxia
PY - 2019/9
Y1 - 2019/9
N2 - Recently, the success of generating fake images by Generative Adversarial Network (GAN) has threatened the authentication of digital images. To address this issue, several automated fake image detectors have been proposed. However, current methods remain vulnerable when testing samples undergo post-processing attacks. In this work, we employed residual signals of chrominance components from multi color spaces, including YCbCr, HSV and Lab, to learn robust deep representations via the well-designed shallow convolutional neural network (CNN). Then, the learned deep representations from different color spaces are concatenated and then fed into the Random Forest (RF), which is the widely used ensemble classifier, to obtain final detection results. Extensive experiments are conducted on the fake image dataset generated by the advanced GAN technique. Experimental results demonstrate the proposed scheme outperforms state-of-the-art methods and achieves the promising average detection accuracy (above 99%) under several post-processing attacks, such as Gaussian blurring and so on.
AB - Recently, the success of generating fake images by Generative Adversarial Network (GAN) has threatened the authentication of digital images. To address this issue, several automated fake image detectors have been proposed. However, current methods remain vulnerable when testing samples undergo post-processing attacks. In this work, we employed residual signals of chrominance components from multi color spaces, including YCbCr, HSV and Lab, to learn robust deep representations via the well-designed shallow convolutional neural network (CNN). Then, the learned deep representations from different color spaces are concatenated and then fed into the Random Forest (RF), which is the widely used ensemble classifier, to obtain final detection results. Extensive experiments are conducted on the fake image dataset generated by the advanced GAN technique. Experimental results demonstrate the proposed scheme outperforms state-of-the-art methods and achieves the promising average detection accuracy (above 99%) under several post-processing attacks, such as Gaussian blurring and so on.
KW - fake image detection
KW - generative adversarial network
KW - multi color spaces
KW - random forest
UR - https://www.scopus.com/pages/publications/85076227818
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85076227818&origin=recordpage
U2 - 10.1109/ICIP.2019.8803740
DO - 10.1109/ICIP.2019.8803740
M3 - RGC 32 - Refereed conference paper (with host publication)
SN - 9781538662496
T3 - Proceedings - International Conference on Image Processing, ICIP
SP - 2299
EP - 2303
BT - 2019 IEEE International Conference on Image Processing - PROCEEDINGS
PB - IEEE Computer Society
T2 - 26th IEEE International Conference on Image Processing (ICIP 2019)
Y2 - 22 September 2019 through 25 September 2019
ER -