Deep learning has become an essential technique in image steganography. Most of the current deep-learning-based steganographic methods process digital images in the spatial domain. There are problems such as limited embedding capacity and unsatisfactory visual quality. To improve capacity-distortion performance, we develop a steganographic method from the frequency-domain perspective. We propose a module called the adaptive frequency-domain channel attention network (AFcaNet), which makes full use of the frequency features in each channel by a fine-grained manner of assigning weights. We apply this module to the state-of-the-art SteganoGAN, forming an Adaptive Frequency High-capacity Steganography Generative Adversarial Network (AFHS-GAN). The proposed neural network enhances the ability of high-dimensional feature extraction through overlaying densely connected convolutional blocks. In addition to this, a low-frequency loss function is introduced as an evaluation metric to guide the training of the network and thus reduces the modification of low-frequency regions of the image. Experimental results on the Div2K dataset show that our method has a better generalization capability compared to the SteganoGAN, with substantial improvement in both embedding capacity and stego-image quality. Furthermore, the embedding distribution of our method in the DCT domain is more similar to that of the traditional method, which is consistent with the prior knowledge of image steganography.

Since 3D models can intuitively display real-world information, there are potential scenarios in many application fields, such as architectural models and medical organ models. However, a 3D model shared through the internet can be easily obtained by an unauthorized user. In order to solve the security problem of 3D model in the cloud, a reversible data hiding method for encrypted 3D model based on prediction error expansion is proposed. In this method, the original 3D model is preprocessed, and the vertex of 3D model is encrypted by using the Paillier cryptosystem. In the cloud, in order to improve accuracy of data extraction, the dyeing method is designed to classify all vertices into the embedded set and the referenced set. After that, secret data is embedded by expanding direction of prediction error with direction vector. The prediction error of the vertex in the embedded set is computed by using the referenced set, and the direction vector is obtained according to the mapping table, which is designed to map several bits to a direction vector. Secret data can be extracted by comparing the angle between the direction of prediction error and direction vector, and the original model can be restored using the referenced set. Experiment results show that compared with the existing data hiding method for encrypted 3D model, the proposed method has higher data hiding capacity, and the accuracy of data extraction have improved. Moreover, the directly decrypted model has less distortion.