Encryption algorithms are usually applied to images in cloud environment to protect privacy, which will make image retrieval difficult. Existing encrypted image retrieval schemes mainly follow the two-step strategy, which firstly extract features from encrypted images and then use these features to conduct retrieval. However, the feature extraction procedure brings additional computational expense, and the extracted feature will result in privacy leakage. In this paper, we propose a new encrypted image retrieval scheme in end-to-end manner. End-to-end means we combine the feature extraction process and the retrieval process, which can reduce computational complexity. Specifically, images are encrypted by DC coefficients encryption, AC coefficients permutation and block permutation. After encryption, the encrypted images can be directly input into our retrieval model, which is based on Vision Transformer (Vit), without any process to conduct retrieval. What’s more, we propose a new type of data augmentation method named random block selection transformation (RST), which is compatible with the Vit backbone and can improve the model performance. Experiment result shows that our scheme can achieve superior retrieval accuracy than other state-to-the-art encrypted image retrieval schemes, while relatively high security can be retained.

The advancement of cloud-assisted Internet of Things (IoT) has amplified the usability for secure and searchable image retrieval, addressing the growing demand for privacy protection in digital multimedia. Current encrypted image retrieval schemes focus either on search precision or search efficiency, making them less viable for deployment on IoT devices with limited resources. Therefore, in this article, we present a privacy-preserving clustering-based image retrieval (PPCBIR) scheme for IoT environment. First, we employ a convolutional neural networks (CNN) for feature extraction and design an extended k-nearest neighbor (kNN) algorithm to protect the privacy of image features. Then, we build a novel clustering-based hierarchical index tree structure to improve retrieval efficiency without compromising data privacy. Subsequently, a matrix re-encryption technique is implemented to achieve the availability of multiterminal key distribution in IoT. Furthermore, we propose an index merging method that is scalable to index trees constructed by different data owners. Finally, formal security analysis demonstrates that PPCBIR is resistant to various threat models. Extensive experiments using authentic datasets indicate that our proposed scheme is comparable to linear retrieval in search accuracy and outperforms existing state-of-the-art schemes in search efficiency, and demonstrate its practicability in IoT.

JPEG image encryption technology converts original images into noise-like images that do not contain any useful information, ensuring the security and privacy of valuable images. At present, most existing work may not achieve a good balance between file size preservation and security. Therefore, in order to solve this issue, we present a new JPEG image encryption scheme. In our scheme, we first perform predictions on the DC coefficients, and then implement averaging operation on the prediction error, which provides security effects while effectively decreases the encoding length. To further strengthen the diffusion performance, block permutation excluding the DC coefficients is executed. In addition, in order to control the file size growth, AC coefficients are protected by the run-based RS (run/size) pairs permutation. The experimental results show that the proposed scheme can not only effectively preserve the image file size, but also offers satisfactory security.

Image retrieval systems help users to browse and search among extensive images in real time. With the rise of cloud computing, retrieval tasks are usually outsourced to cloud servers. However, the cloud scenario brings a daunting challenge of privacy protection as cloud servers cannot be fully trusted. To this end, image-encryption-based privacy-preserving image retrieval (PPIR) schemes have been developed, which first extract features from cipher-images, and then build retrieval models based on these features. Yet, most existing PPIR approaches extract shallow features and design trivial unsupervised retrieval models, resulting in insufficient expressiveness for the cipher-images. In this paper, we propose a novel paradigm named Encrypted Vision Transformer (EViT), which advances the discriminative representations capability of cipher-images. First, to capture comprehensive ruled information, we extract multi-level local length sequence and global Huffman-Code frequency features from the cipher-images which are encrypted by permutation encryption, sign encryption, and stream cipher during the JPEG compression process. Second, we design the modified self-supervised Vision Transformer with Huffman-embedding and propose two robust data augmentations on cipher-images to improve representation power of the retrieval model. Moreover, our proposal can be easily adapted to unsupervised or supervised settings. Extensive experiments reveal that EViT achieves both excellent encryption and retrieval performance, outperforming current schemes in terms of retrieval accuracy by large margins while protecting image privacy effectively. Code is publicly available at https://github.com/onlinehuazai/EViT.

In recent years, people tend to upload images to cloud servers, which provide storage and retrieval functions. To prevent users’ privacy from leaking to the server, research on cipher-image retrieval has attracted much attention. This work presents a novel encrypted image retrieval method. With this scheme, we perform encryption during the JPEG compression process by applying 16×16 DCT (Discrete Cosine Transform) for blocks’ transformation, followed by coefficients distribution and 8×8 blocks’ permutation. For the retrieval part, when an encrypted query image is sent by an authorized user, the server extracts its DCT histograms as features and inputs them into our trained network model, which incorporates transpose Multilayer perceptron modules ($Transpose$$MLP$), for retrieval. Experimental results show that our scheme, compared with related schemes, can improve the retrieval performance significantly, when ensuring compression friendliness and no feature information leakage. Moreover, our scheme enables cipher-image retrieval from multiple image owners.

Joint image compression and encryption techniques can be used to ensure the security of JPEG images. The related works in this field face problems such as low compression efficiency, limited protection power, and corrupted format information. In this paper, we propose a new JPEG protection method by introducing encryption operations into the compression process. Considering that different images have different properties, instead of using a fixed block size for encryption, we develop a new block segmentation strategy to first divide the plain-image into two different sizes of blocks,$8\times 8$and$16\times 16$, followed by corresponding size of DCT’s transformation. The strategy can ensure that the file size increment and rate-distortion are minimal. After block transformation and quantization, we encrypt DC coefficients by XOR and swapping, while AC coefficients are protected through block permutation and data embedding. The experiments show that our scheme does not compromise JPEG’s compression efficiency, offers satisfied security and maintains the format-compliant of the final encrypted bitstream to JPEG’s decoder.

Applying encryption technology to image retrieval can ensure the security and privacy of personal images. The related researches in this field have focused on the organic combination of encryption algorithm and artificial feature extraction. Many existing encrypted image retrieval schemes cannot prevent feature leakage and file size increase or cannot achieve satisfied retrieval performance. In this paper, a new end-to-end encrypted image retrieval scheme is presented. First, images are encrypted by using block rotation, new orthogonal transforms and block permutation during the JPEG compression process. Second, we combine the triplet loss and the cross entropy loss to train a network model, which contains gMLP modules, by end-to-end learning for extracting cipher-images' features. Compared with manual features extraction such as extracting color histogram, the end-to-end mechanism can economize on manpower. Experimental results show that our scheme has good retrieval performance, while can ensure compression friendly and no feature leakage.

This paper studies Physical-layer Network Coding (PNC) in a two-way relay channel (TWRC) operated based on OFDM and QPSK modulation but with the presence of carrier frequency offset (CFO). CFO, induced by node motion and/or oscillator mismatch, causes inter-carrier interference (ICI) that impairs received signals in PNC. Our ultimate goal is to empower the relay in TWRC to decode network-coded information of the end users at a low bit error rate (BER) under CFO, as it is impossible to eliminate the CFO of both end users. For that, we first put forth two signal detection and channel decoding schemes at the relay in PNC. For signal detection, both schemes exploit the signal structure introduced by ICI, but they aim for different output, thus differing in the subsequent channel decoding. We then consider CFO compensation that adjusts the CFO values of the end nodes simultaneously and find that an optimal choice is to yield opposite CFO values in PNC. Particularly, we reveal that pilot insertion could play an important role against the CFO effect, indicating that we may trade more pilots for not just a better channel estimation but also a lower BER at the relay in PNC. With our proposed measures, we conduct simulation using repeat-accumulate (RA) codes and QPSK modulation to show that PNC can achieve a BER at the relay comparable to that of point-to-point transmissions for low to medium CFO levels.

In recent years, image encryption has been broadly researched. Since a large proportion of images on Internet are compressed, and JPEG is the most widely adopted standards for image compression, a variety of joint image compression and encryption algorithms have been proposed. JPEG image encryption uses signal processing techniques as well as cryptographic techniques in different stages of the JPEG compression process. Based on the location where the encryption is taken place, these algorithms can be generally classified into three major categories: pre‐compression encryption algorithms, in‐compression encryption algorithms, and post‐compression encryption algorithms, which correspond to conduct encryption operations before, during, and after the compression process. From this perspective of classification, the authors give a comprehensive survey on representative image encryption algorithms of each type, and show their properties and limitations. Some of the most recent encryption schemes that achieve protection at various positions of the compression process are selected for comparison, which are two pre‐compression encryption algorithms, four in‐compression encryption algorithms, and two post‐compression encryption algorithms. Possible future research directions on designing joint JPEG compression and encryption schemes are provided in the end, which may facilitate solving application scenario‐oriented JPEG security problems with new technologies.

For joint image compression and encryption schemes, the encryption power and compression efficiency are commonly two contradictory things. In this paper, we propose a new joint image compression and encryption scheme based on a lossy JPEG standard, which aims at encryption power's enhancement, on the premise of maintaining the JPEG's compression efficiency. The proposed scheme is image-content-adaptive, since the secret encryption key is generated from the plain image using the BLAKE2 hash algorithm. Three encryption operations are contained in our scheme, alternating new orthogonal transforms transformation, dc coefficients encryption, and ac coefficients encryption. To save the cost for transmitting different encryption keys each time to decoder for decryption when the plain image changes, we propose embedding the encryption key into the entropy-encoded bitstream of some ac coefficients, and the whole embedding procedure is controlled by another secret key called the embedding key. Extensive experiments are conducted to show that our encryption scheme is JPEG friendly, and has good confusion and diffusion properties. Detailed security analysis is also given to illustrate the proposed scheme's persistence to various cryptanalysis strategies.

In this paper, we propose a novel joint image compression and encryption technique where an annoying image can still be recovered even without the encryption key. Our work is based on JPEG standard. By embedding encryption algorithm at the transformation stage, we realize image encryption and compression together with controllable image quality. Instead of using the 8×8 discrete cosine transform (DCT) alone for transformation, we develop new orthogonal transforms by introducing sign-flips into the butterflies of DCT's flow-graph structure, and then employ them alternatively in JPEG's transformation stage according to a secret key. By carefully selecting the butterflies for sign-flipping, we can control the visual quality of the encrypted images. Finally, a detailed security analysis of our proposed encryption algorithm is presented to show its resistance to various attacks, such as cryptographic attack, replacement attack and statistical model-based attack.