In this work, we propose to use deep learning to segment an image based on its color and its content. We start by using the content-color-dependent screening (CCDS) developed previously in [1]. The goal of CCDS is to apply different color assignments for the two or more regular or irregular halftones within the image depending on the local color and content of the image. If the image content contains high variance of color and texture locally, the artifacts due to halftoning will not be as visible as the artifacts in smooth areas of the image [1]. Therefore, the goal of CCDS was to detect smooth areas of the image and apply the best color assignments to those areas. In order to detect the smooth areas, the image segmentation algorithm involving the retrieval of the cluster-map and the segmented edge-map was proposed [1]. The main drawback of the proposed approach is that for a given image, the result highly depends on the initial parameters, such as the number of clusters, low and high thresholds for edge detection, bilateral filter parameters and others. In this work, we propose to use the well-known U-net architecture to detect the smooth areas of the image. U-net is a type of a convolutional neural network (CNN) designed for fast, accurate image segmentation, and it is used to predict a label for every single pixel [2]. The architecture of the U-net is suitable for this work because it consists of a contracting path to capture context and a symmetrical expansive path that enables precise localization [2]. We believe that using the U-net to detect smooth areas of the image will greatly improve the current approach and provide better results.
Realistic image forgeries involve a combination of splicing, resampling, cloning, region removal and other methods. While resampling detection algorithms are effective in detecting splicing and resampling, copy-move detection algorithms excel in detecting cloning and region removal. In this paper, we combine these complementary approaches in a way that boosts the overall accuracy of image manipulation detection. We use the copy-move detection method as a pre-filtering step and pass those images that are classified as untampered to a deep learning based resampling detection framework. Experimental results on various datasets including the 2017 NIST Nimble Challenge Evaluation dataset comprising nearly 10,000 pristine and tampered images shows that there is a consistent increase of 8%-10% in detection rates, when copy-move algorithm is combined with different resampling detection algorithms.
The amount of digital imagery recorded has recently grown exponentially, and with the advancement of software, such as Photoshop or Gimp, it has become easier to manipulate images. However, most images on the internet have not been manipulated and any automated manipulation detection algorithm must carefully control the false alarm rate. In this paper we discuss a method to automatically detect local resampling using deep learning while controlling the false alarm rate using a-contrario analysis. The automated procedure consists of three primary steps. First, resampling features are calculated for image blocks. A deep learning classifier is then used to generate a heatmap that indicates if the image block has been resampled. We expect some of these blocks to be falsely identified as resampled. We use a-contrario hypothesis testing to both identify if the patterns of the manipulated blocks indicate if the image has been tampered with and to localize the manipulation. We demonstrate that this strategy is effective in indicating if an image has been manipulated and localizing the manipulations.