Wei Zhang

To enhance the security and transmission efficiency of facial images, this paper proposes a novel Multi-Coupled Map Lattice with Dynamic Delayed Feedback (MCML-DDF) spatiotemporal chaotic system and its application in facial image compressive encryption. The MCML-DDF system is designed to generate highly secure key streams, forming the foundation of the encryption algorithm. Performance analysis demonstrates that the Kolmogorov–Sinai Entropy Breadth (KEB) increases from 82.4675% to 100%, and the Kolmogorov–Sinai Entropy Density (KED) rises from 0.289455 to 7.8766924, confirming the expanded parameter. During encryption, the facial region is accurately identified and encrypted to protect the critical region, while the background is compressed via compressed sensing driven by the chaotic measurement matrix. For image scrambling, a CMT-based Dynamic Bidirectional Randomized Scrambling (CMT-DBRS) algorithm ensures uniform and unpredictable pixel permutation. For image diffusion, a Dynamic Kernel Diffusion (DKD) algorithm improves diffusion efficiency and resistance against attacks. Experimental results confirm that the proposed algorithm achieves secure and efficient facial image encryption and further validate the improved chaotic characteristics and security of the MCML-DDF chaotic system.

In recent years, most proposed digital image encryption algorithms have primarily focused on encrypting raw pixel data, often neglecting the integration with image compression techniques. Image compression algorithms, such as JPEG, are widely utilized in internet applications, highlighting the need for encryption methods that are compatible with compression processes. This study introduces an innovative color image encryption algorithm integrated with JPEG compression, designed to enhance the security of images susceptible to attacks or tampering during prolonged transmission. The research addresses critical challenges in achieving an optimal balance between encryption security and compression efficiency. The proposed encryption algorithm is structured around three key compression phases: Discrete Cosine Transform (DCT), quantization, and entropy coding. At each stage, the algorithm incorporates advanced techniques such as block segmentation, block replacement, DC coefficient confusion, non-zero AC coefficient transformation, and RSV (Run/Size and Value) pair recombination. Extensive simulations and security analyses demonstrate that the proposed algorithm exhibits strong robustness against noise interference and data loss, effectively meeting stringent security performance requirements.

Recent advances in text-to-image synthesis have captivated audiences worldwide, drawing considerable attention. Although significant progress in generating photo-realistic images through large pre-trained autoregressive and diffusion models, these models face three critical constraints: (1) The requirement for extensive training data and numerous model parameters; (2) Inefficient, multi-step image generation process; and (3) Difficulties in controlling the output visual features, requiring complexly designed prompts to ensure text-image alignment. Addressing these challenges, we introduce the CLIP-GAN model, which innovatively integrates the pretrained CLIP model into both the generator and discriminator of the GAN. Our architecture includes a CLIP-based generator that employs visual concepts derived from CLIP through text prompts in a feature adapter module. We also propose a CLIP-based discriminator, utilizing CLIP's advanced scene understanding capabilities for more precise image quality evaluation. Additionally, our generator applies visual concepts from CLIP via the Text-based Generator Block (TG-Block) and the Polarized Feature Fusion Module (PFFM) enabling better fusion of text and image semantic information. This integration within the generator and discriminator enhances training efficiency, enabling our model to achieve evaluation results not inferior to large pre-trained autoregressive and diffusion models, but with a $94\%$ reduction in learnable parameters. CLIP-GAN aims to achieve the best efficiency-accuracy trade-off in image generation given the limited resource budget. Extensive evaluations validate the superior performance of the model, demonstrating faster image generation speed and the potential for greater stylistic diversity within the GAN model, while still preserving its smooth latent space.