A high-capacity coverless image steganography method based on double-level index and block matching.

Recently, a new information hiding technology called coverless information steganography (CIS) is proposed, which uses the original natural image as stego image for the transmission of secret information which can resist the detection of image steganalysis algorithm, so it received extensive attention and support. However, it is still a low hidden capacity of the CIS methods up to now. This paper proposes a high-capacity coverless information steganography technology. In which we divide the cover image into several image blocks and every image block can represent one bit secret information which improves the capacity greatly. Then we retrieve the image blocks for replacement from the image block database based on secret information and then synthesize them into stego image. The quality of the stego image is still high because the required image blocks are similar to cover image blocks and they are all from natural images. Moreover, in order to improve the retrieval efficiency, we have established a double-level index structure. The experimental results show that compared with the existing CIS methods, the proposed method has larger capacity and better visual quality.

Information hiding based on Augmented Reality.

Information hiding aims to achieve secret communication via certain carrier. However, these carrier-based methods often have different kinds of deficiencies. In order to solve the problems addressed by the traditional information hiding methods such as the difficult balance between secret embedding rate and detection rate, this paper proposes a novel approach which utilizes Augmented Reality (AR) to achieve secret communication. In this paper, we present an AR based information hiding architecture which combines information hiding, augmented reality, and deep learning methods altogether. The proposed architecture basically follows the idea of secret-key matching policy. The secret sender first maps the secret message to objects, images or coordinates, etc. The mapped objects, images or coordinates then serve as the secret key for further secret revealing. The secret key and concealing model are shared between two communication parties instead of direct transmitting the secret messages. Different secret keys can be combined in order to generate more mapping sequences. Also, deep learning based models are integrated in the architecture to extend the mapping varieties. By taking advantage of the augmented reality technique, the secret messages can be transmitted in various formats which results in higher secret embedding rate in potential. Furthermore, the proposed architecture can be seen as a useful application of coverless information hiding scheme. The experimental system realizes the proposed architecture by implementing convolutional neural network (CNN) based real-time object detection, image recognition, augmented reality and secret-key matching altogether which shows great promise in practice.

An anti-forensic scheme on computer graphic images and natural images using generative adversarial networks.

Computer graphic images (CGI) can be manufactured very similar to natural images (NI) by state-of-the-art algorithms in computer graphic filed. Thus, there are various identification algorithms proposed to detect CGI. However, the manipulation is complicated and difficult for an ultimate CGI against the forensic algorithms. Further, the forensics on CGI and NI made achievements in the different aspects with the encouragement of deep learning. Though the generated CGI can achieve high quality automatically by generative adversarial networks (GAN), CGI generation based on GAN is difficult to ensure that it cannot be detected by forensics. In this paper, we propose a brief and effective architecture based on GAN for preventing the generated images being detected under the forensics on CGI and NI. The adapted characteristics will make the CGI generated by GAN fools the detector and keep the end-to-end generation mode of GAN.

Airline Passenger Profiling Based on Fuzzy Deep Machine Learning.

Passenger profiling plays a vital part of commercial aviation security, but classical methods become very inefficient in handling the rapidly increasing amounts of electronic records. This paper proposes a deep learning approach to passenger profiling. The center of our approach is a Pythagorean fuzzy deep Boltzmann machine (PFDBM), whose parameters are expressed by Pythagorean fuzzy numbers such that each neuron can learn how a feature affects the production of the correct output from both the positive and negative sides. We propose a hybrid algorithm combining a gradient-based method and an evolutionary algorithm for training the PFDBM. Based on the novel learning model, we develop a deep neural network (DNN) for classifying normal passengers and potential attackers, and further develop an integrated DNN for identifying group attackers whose individual features are insufficient to reveal the abnormality. Experiments on data sets from Air China show that our approach provides much higher learning ability and classification accuracy than existing profilers. It is expected that the fuzzy deep learning approach can be adapted for a variety of complex pattern analysis tasks.