Median filtering is used as an anti-forensic technique to erase processing history of some image manipulations such as JPEG, resampling, etc. Thus, various detectors have been proposed to detect median filtered images. To counter these techniques, several anti-forensic methods have been devised as well. However, restoring the median filtered image is a typical ill-posed problem, and therefore it is still difficult to reconstruct the image visually close to the original image. Also, it is further hard to make the restored image have the statistical characteristic of the raw image for the anti-forensic purpose. To solve this problem, we present a median filtering anti-forensic method based on deep convolutional neural networks (CNNs), which can effectively remove traces from median filtered images. We adopt the framework of generative adversarial networks (GANs) to generate images that follow the underlying statistics of unaltered images, significantly enhancing forensic undetectability. Through extensive experiments, we demonstrate that our method successfully deceives the existing median filtering forensic techniques.

We herein advance a secure steganographic algorithm that uses a turnover policy and a novel adjusting process. Although the method of Wang et al. uses Pixel-Value Differencing (PVD) and their modulus function provides high capacity and good image quality, the embedding process causes a number of artifacts, such as abnormal increases and fluctuations in the PVD histogram, which may reveal the existence of the hidden message. In order to enhance the security of the algorithm, a turnover policy is used that prevents abnormal increases in the histogram values and a novel adjusting process is devised to remove the fluctuations at the border of the subrange in the PVD histogram. The proposed method therefore eliminates all the weaknesses of the PVD steganographic methods thus far proposed and guarantees secure communication. In the experiments described herein, the proposed algorithm is compared with other PVD steganographic algorithms by using well-known steganalysis techniques, such as RS-analysis, steganalysis for LSB matching, and histogram-based attacks. The results support our contention that the proposed method enhances security by keeping the PVD histogram similar to the cover, while also providing high embedding capacity and good imperceptibility to the naked eye.