Zero-watermarking is an effective solution for image copyright protection without altering the original content. However, current deep learning-based methods suffer from two key limitations. First, most feature extraction networks, originally designed for classification, lack robust feature learning essential for resisting attacks. Second, conventional methods seldom incorporate the generated watermark back into training, missing opportunities to further optimize the model. To address these issues, we propose ConZWNet, a two-stage framework that integrates contrastive learning with feedback-driven zero-watermark generation. In the first stage, we use ConvNeXt to learn invariant, attack-resistant features via contrastive learning on weak–strong augmentation. In the second stage, a residual network coupled with a Multi-Layer Perceptron (MLP) fuses features from host and copyright images to produce a latent zero-watermark, which is then verified by an MLP-based copyright identification network. This feedback loop optimizes feature fusion and transforms zero-watermark generation into a self-supervised process. Extensive experiments demonstrate that ConZWNet achieves state-of-the-art robustness against various attacks while ensuring high distinguishability among host images and copyrights. Ablation studies confirm the effectiveness of components, including two-stage architecture, contrastive learning, weak–strong augmentation, and copyright identification network. The source code is publicly available at https://github.com/hanhongxin1028/ConZWNet .

The rapid development of deepfake technology poses challenges to face-centered data security. Existing methods primarily focus on how to transfer deepfake detectors from the source domain to the target domain to handle diverse deepfake techniques. In practical application scenarios, it is usually difficult to access the true and false labels of the source domain. In this letter, we introduce a new adaptation framework called Latent Domain Knowledge Distillation (LDKD) for cross-domain deepfake detection. In the proposed framework, we construct a knowledge distillation structure that includes a student network and a teacher network, which are jointly optimized in a coupled manner to facilitate the model's adaptation to the target domain. Furthermore, to improve the quality of pseudo-labels generated by the teacher network, we propose a Fourier Latent Domain Generation Module (FLGM) and a Stochastic Complementary Mask Module (SCMM). The former is used to generate latent domains to bridge domain differences at the image level, while the latter is employed to mine richer contextual cues for the model. Extensive cross-domain experimental results demonstrate that our method achieves state-of-the-art performance, and the model analysis proves the effectiveness of our key components.

The current remote sensing image data sharing frameworks confront dual challenges of trusted transaction and copyright protection. On the one hand, existing zero-watermark schemes are constrained by the hand-crafted feature extraction paradigms, which struggle to capture the intricate high-dimensional spectral-spatial joint feature representations inherent in remote sensing imagery. On the other hand, most data trading protocols fail to harmonize fairness in transactions with robust copyright protection. To this end, a zero-watermark method based on multi-channel PCNN and blockchain for remote sensing image transaction certificate and copyright protection is proposed. The method designs a multi-channel PCNN model to process each band of remote sensing images in parallel, and uses multi-scale morphological gradient and differential box counting to optimize and motivate the model to enhance the robustness of zero-watermark. The distinguishing performance of zero-watermarks for copyright information is enhanced according to the dynamic features and spatio-temporal characteristics of PCNN time series. In addition, given the characteristics of remote sensing image data volume is huge and involves private and sensitive information, a zero-watermark registration and extraction framework integrating Hyperledger Fabric and Inter-Planetary File System is constructed, aiming to achieve the purpose of transaction verification and copyright protection.

For ensuring the integrity of high-resolution remote sensing (HRRS) images, the perceptual hash method offers a dual advantage: it preserves the non-destructive nature of the original image while also ensuring robustness to content-preserving operations. However, current deep learning based HRRS image hashing methods for integrity authentication are notably limited as they terminate at the feature extraction stage and fail to achieve an end-to-end construction from image to hash value. Consequently, there is a looming risk of uncontrollability and unexpected events. To overcome this problem, this paper proposes A Deep Subject-Sensitive Hashing Network (DSSHN), presenting a unified network for end-to-end feature extraction and hash construction. Improved Convolutional Block Attention Module (I-CBAM) helps the network to focus more on subject-sensitive features. A targeted training scheme ensures perceptual hash robustness. Experimental results reveal that the algorithm achieves the best tampering detection performance, with top AUC (0.994) and leading precision and recall rates.

This paper is to study the control scheme for neutral-point-clamping three-level (NPC-3L) inverters fed dual three-phase permanent magnet synchronous motor (PMSM) drives based on vector space decomposition (VSD). The key is to propose a novel VSD based space vector modulation (SVM) strategy for the dual-three phase drives. The torque of the PMSM drive is controlled by voltage components on α-β plane while the current harmonics are suppressed by voltage components on x - y plane of the VSD-SVM. The upper and lower capacitor voltages are balanced well in DC link. Based on the proposed VSD-SVM strategy, a simple control scheme is presented for the dual three-phase PMSM drives. The principles of the proposed modulation strategy and control scheme are described in detail. The experiments are carried on a laboratory prototype to verify that the proposed VSD-SVM based control scheme can provide both good steady-state and dynamic performance for the dual three-phase PMSM drives.