Defeating data hiding in social networks using generative adversarial network

Abstract

As a large number of images are transmitted through social networks every moment, terrorists may hide data into images to convey secret data. Various types of images are mixed up in the social networks, and it is difficult for the servers of social networks to detect whether the images are clean. To prevent the illegal communication, this paper proposes a method of defeating data hiding by removing the secret data without impacting the original media content. The method separates the clean images from illegal images using the generative adversarial network (GAN), in which a deep residual network is used as a generator. Therefore, hidden data can be removed and the quality of the processed images can be well maintained. Experimental results show that the proposed method can prevent secret transmission effectively and preserve the processed images with high quality.

Destroy the Robust Commercial Watermark via Deep Convolutional Encoder-Decoder Network

“…In this section, we will show the results of attack on watermarking of different strengths through our method and the comparison experiments with Wang, Qian, Feng, and Zhang [26] and StirMark. We choose the "F16" in Set9 as the test image and embed the watermark with the strength a with 1, 2, 3, and 4, and then, we will get the four images to be attacked.…”

“…We choose the "F16" in Set9 as the test image and embed the watermark with the strength a with 1, 2, 3, and 4, and then, we will get the four images to be attacked. Table 1 shows the results after processing with the method proposed, and the comparison with Wang et al [26] and several common geometric attacks in StirMark. Let the parameter of the cropping attack be c, and the value is 0.75, 0.5, 0.25 { }; the rotation attack parameter be r, and the value is 180 °, 150 °, 90 °{ }; and the scaling attack parameter be sc, and the value is 0.5, 1.5, 2 { }.…”

“…Only the effect of the attack makes the software identification result "none," and this attack is meaningful. Although the method proposed in Wang et al [26] can also meet the requirement, the proposed method is more convenient when training. In our opinion, the visual quality of the image should not be affected too much while the watermark is erased.…”