IEEE Transactions on Circuits and Systems for Video Technology最新文献_第3页

AdaGuard: Location-Aware Adaptive Defense Against Adversarial Patch Attacks in Vision Transformers AdaGuard：针对视觉变形器中对抗性补丁攻击的位置感知自适应防御

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-06 DOI: 10.1109/TCSVT.2026.3710546

Jinyin Hu;Jiawei Zhou;Zhonghao Yang;Huadi Zheng;Daojing He;Yu Li

{"title":"AdaGuard: Location-Aware Adaptive Defense Against Adversarial Patch Attacks in Vision Transformers","authors":"Jinyin Hu;Jiawei Zhou;Zhonghao Yang;Huadi Zheng;Daojing He;Yu Li","doi":"10.1109/TCSVT.2026.3710546","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3710546","url":null,"abstract":"Vision Transformers (ViTs) have achieved state-of-the-art performance across a wide range of vision tasks, yet they remain highly vulnerable to adversarial patch attacks—deliberate image modifications confined to small, localized regions (patches) that can completely mislead the model. Existing defenses often treat all adversarial inputs uniformly, which either removes essential features or fails to eliminate malicious influence, thereby limiting robustness. We present AdaGuard, an adaptive defense framework that combines adversarial patch detection with location-aware treatment. AdaGuard first employs a lightweight, model-agnostic detector based on patch–neighbor discontinuity to accurately localize adversarial regions of unknown size and position. It then distinguishes between <italic>non-catastrophic</i> attacks (patches in less informative regions) and <italic>catastrophic</i> attacks (patches overlapping with essential features). For non-catastrophic cases, AdaGuard reconstructs the region with neighboring content to fully remove adversarial information, while for catastrophic cases it applies fine-grained attention suppression to neutralize adversarial influence without discarding critical features. Experiments on 12 ViT models and multiple attack methods on ImageNet demonstrate that AdaGuard achieves state-of-the-art robust accuracy while preserving clean accuracy, offering a practical defense for ViTs in real-world applications.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15305-15316"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097103","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Priors Meet Hindsight: An Asymmetric Collaborative Evolution Framework for RGB-D Road Defect Detection 前见后见：RGB-D道路缺陷检测的非对称协同进化框架

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-05 DOI: 10.1109/TCSVT.2026.3700185

Wujie Zhou;Zijun Ju;Liang Liao;Runmin Cong

{"title":"Priors Meet Hindsight: An Asymmetric Collaborative Evolution Framework for RGB-D Road Defect Detection","authors":"Wujie Zhou;Zijun Ju;Liang Liao;Runmin Cong","doi":"10.1109/TCSVT.2026.3700185","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3700185","url":null,"abstract":"Road defect detection represents a fundamental perceptual task for mobile robots and autonomous driving systems; however, existing approaches struggle to balance detection accuracy and computational efficiency on resource-constrained platforms. To overcome this limitation, we propose ACEF, an asymmetric collaborative evolution framework that addresses two fundamental gaps: 1) depth-guided distillation methods tightly couple geometric enhancement with multimodal overhead at inference time and 2) temporal knowledge transfer methods rely solely on correlation-based feature alignment across historical states. ACEF establishes a collaborative learning mechanism between a computationally intensive RGB-D teacher network and an efficient RGB student network, incorporating two key innovations: 1) geometry-aware hash injection, which encodes depth features as binary hash attention masks through discrete cosine transform-based frequency-domain decomposition, injecting geometric priors only during training while imposing no overhead at inference time; and 2) counterfactual trajectory retrospection, which grounds recurrent knowledge transfer within a formal structural causal framework. By formalizing causal validity filtering, ACEF retains only those historical states that empirically reduce the geometric learning gap between the teacher and the student, thereby enabling causally selective, rather than purely correlation-based, trajectory guidance. Comprehensive experimental validation demonstrates that ACEF attains an inference speed of 100.8 FPS in a lightweight unimodal setting, outperforming 19 state-of-the-art methods in detection accuracy while maintaining computational practicality for real-world robotic deployment. Additional evaluations across three diverse datasets further verify the generalizability and adaptability of the proposed framework in robotic operating environments. The code is publicly available at <uri>https://github.com/MuXin99/ACEF</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14882-14896"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097106","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Compressed Video Quality Enhancement Based on Optimal Reference Selection 基于最优参考选择的压缩视频质量增强

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-24 DOI: 10.1109/TCSVT.2026.3706826

Guangming Ren;Jielian Lin;Liqun Lin;Xu Wang;Tiesong Zhao;Xiaoqiang Lu

{"title":"Compressed Video Quality Enhancement Based on Optimal Reference Selection","authors":"Guangming Ren;Jielian Lin;Liqun Lin;Xu Wang;Tiesong Zhao;Xiaoqiang Lu","doi":"10.1109/TCSVT.2026.3706826","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3706826","url":null,"abstract":"Compressed video enhancement (CVE) leverages both inter-frame information and codec bitstream cues, but doing so effectively requires choosing references that align with the prediction structure. In contrast, most existing methods simply stack neighboring frames, which often yields redundant references and makes it difficult to control error propagation along prediction chains. We address this gap with Optimal Reference Selection (ORS), which selects a single reference for each target frame while explicitly accounting for the codec prediction structure via prediction-path independence, low temporal-layer preference, and temporal proximity. Based on ORS, we present ORS-CVE, which combines the selected reference with bitstream priors for enhancement. Our model performs robust long-range alignment through motion-vector guided hierarchical optical flow refinement, and it restores textures and fine details using low-QP residual cues from the lowest temporal layer. Experiments demonstrate consistent improvements in both quantitative metrics and visual quality over prior CVE approaches. Source code available at <uri>https://github.com/RenGM123/ORSCVE</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15409-15420"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

State-Aware Graphical Model for Event Camera Denoising via Joint Maximum a Posteriori Inference 基于联合极大后验推理的事件相机去噪状态感知图形模型

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-03 DOI: 10.1109/TCSVT.2026.3709988

Wentao Wu;Qingchuan Li;Yiyi Zhang;Cheng Chen;Peng Guo

{"title":"State-Aware Graphical Model for Event Camera Denoising via Joint Maximum a Posteriori Inference","authors":"Wentao Wu;Qingchuan Li;Yiyi Zhang;Cheng Chen;Peng Guo","doi":"10.1109/TCSVT.2026.3709988","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709988","url":null,"abstract":"In high-speed scenes, conventional cameras suffer from severe motion blur. Event cameras naturally capture sharp edges and generate high-temporal-resolution event streams, yet their inherent noise severely impairs imaging quality. Existing denoising methods either rely on hand-crafted spatio-temporal rules that are sensitive to scene dynamics and threshold design or rely on deep models with limited generalization and high computational cost. We propose the State-Aware Graphical Event Denoiser (SAGE), a probabilistic framework that couples event-wise local Kalman state estimation with a state-conditioned graphical model to perform joint maximum a posteriori inference over event labels and latent states. SAGE explicitly models event activity, temporal dynamics, and state uncertainty, and incorporates innovation-based adaptive covariance estimation to handle heterogeneous noise. It further employs density- and motion-adaptive neighborhoods and robust potentials to suppress outliers and spurious triggers. The resulting objective is efficiently solved via alternating minimization. Extensive experiments on multiple public datasets show that SAGE consistently outperforms existing denoising methods across multiple evaluation metrics, yielding more than 12% improvement while better preserving structural details and event fidelity. These results indicate that combining uncertainty-aware state estimation with structured probabilistic inference offers an effective and practical solution for event-stream denoising. The code is publicly available at <uri>https://github.com/chengchen512/SAGE</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15462-15477"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IOTA: Corrective Knowledge-Guided Prompt Learning via Black-White Box Framework 基于黑白盒框架的纠正性知识引导快速学习

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-24 DOI: 10.1109/TCSVT.2026.3706761

Shaokun Wang;Yifan Yu;Yuhang He;Weili Guan;Yihong Gong

{"title":"IOTA: Corrective Knowledge-Guided Prompt Learning via Black-White Box Framework","authors":"Shaokun Wang;Yifan Yu;Yuhang He;Weili Guan;Yihong Gong","doi":"10.1109/TCSVT.2026.3706761","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3706761","url":null,"abstract":"Recently, adapting pre-trained models to downstream tasks has attracted increasing interest. Previous Parameter-Efficient-Tuning (PET) methods regard the pre-trained model as an opaque Black Box model, relying purely on data-driven optimization and underutilizing their inherent prior knowledge. This oversight limits the models’ potential for effective downstream task adaptation. To address these issues, we propose a novel black-whIte bOx prompT leArning framework (IOTA), which integrates a data-driven Black Box module with a knowledge-driven White Box module for downstream task adaptation. Specifically, the White Box module derives corrective knowledge by contrasting the wrong predictions with the right cognition. This knowledge is verbalized into interpretable human prompts and leveraged through a corrective knowledge-guided prompt selection strategy to guide the Black Box module toward more accurate predictions. By jointly leveraging knowledge- and data-driven learning signals, IOTA achieves effective downstream task adaptation. Experimental results across multiple downstream benchmarks and tasks under few-shot and easy-to-hard adaptation settings demonstrate the effectiveness of corrective knowledge and the superiority of our method over state-of-the-art methods.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15082-15096"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IRANet: Illumination-Aware Dual-Branch Reversible Aggregation Network for Low-Light Image Enhancement 用于弱光图像增强的照度感知双分支可逆聚合网络

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-03 DOI: 10.1109/TCSVT.2026.3709887

Mengmeng Zhang;Zihao Yu;Huihui Bai;Hongyuan Jing;Zhi Liu

{"title":"IRANet: Illumination-Aware Dual-Branch Reversible Aggregation Network for Low-Light Image Enhancement","authors":"Mengmeng Zhang;Zihao Yu;Huihui Bai;Hongyuan Jing;Zhi Liu","doi":"10.1109/TCSVT.2026.3709887","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709887","url":null,"abstract":"LLIE is challenged by illumination entanglement in the HVI color space, where coupling between illumination and chromaticity yields mismatched color–structure recovery. We propose the illumination-aware Dual-Branch Reversible Aggregation Network (IRANet), which reformulates enhancement as illumination-aware reasoning in HVI. However, under low-light conditions, the lack of long-distance contextual relationships compromises the stability of illumination-aware inference. To address the issue, we propose the Axis-Reversible Context Aggregator (ARCA), which restores long-distance context and stabilizes feature aggregation in both branches. The coherence of illumination reasoning is further hindered by limited synergy between the illumination and chromaticity branches. The Bidirectional illumination-aware Dependency (BIAD) mechanism establishes bidirectional illumination-aware dependencies between the I and HV branches, reinforcing coordinated fusion of illumination and chromaticity information. In addition, low-light enhancement still suffers from the lack of global consistency and local fidelity, for which the Multi-Scale Spatial–Channel Gating (MSCG) mechanism refines illumination and chromaticity information through spatial–channel gating and multi-scale channel gating to jointly maintain global consistency and local structural fidelity. Extensive experiments demonstrate that IRANet delivers SOTA performance with natural color restoration and structural fidelity under diverse low-light conditions.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15448-15461"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097315","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-Level Explicit Feature Alignment Network for Unsupervised Domain Adaptive Person Search 无监督域自适应人物搜索的多级显式特征对齐网络

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-17 DOI: 10.1109/TCSVT.2026.3704656

Xi Yang;Hexun Zhou;De Cheng;Menghui Tian;Nannan Wang

{"title":"Multi-Level Explicit Feature Alignment Network for Unsupervised Domain Adaptive Person Search","authors":"Xi Yang;Hexun Zhou;De Cheng;Menghui Tian;Nannan Wang","doi":"10.1109/TCSVT.2026.3704656","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3704656","url":null,"abstract":"Unsupervised domain adaptive person search faces significant challenges due to distribution differences between domains. Existing methods primarily rely on implicit alignment strategies, which often result in low accuracy and severe label noise when confronted with intra-domain scale variations and pseudo-label noise accumulation. To address these challenges, we propose a novel Multi-level Explicit Feature Alignment Network (MEFANet), which achieves fine-grained cross-domain alignment through explicit feature manipulation from three perspectives: sample-level distribution, semantic-level scale, and instance-level label. Specifically, we design a Dynamic Inter-Domain Balance (DIDB) module that partitions samples according to domain classifier confidence, facilitating cross-domain knowledge transfer for heterogeneous samples while enforcing domain-invariant alignment for homogeneous ones to calibrate sample-level distributions. To tackle individual scale diversity, we develop a Hierarchical Frequency Decoupling (HFD) module that leverages spectral analysis to fuse shallow high-frequency edge features with deep low-frequency semantic features. This frequency-aware explicit fusion strategy effectively enhances the representation of pedestrians across varying distances and resolutions. Furthermore, we establish a Bidirectional Feedback Collaborative Optimization (BiFCO) mechanism that iteratively refines pseudo-labels through dynamic interaction with re-identification (ReID) tasks and quality assessment based on joint feature similarity and spatial consistency, thereby enhancing their reliability. Experimental results demonstrate that our MEFANet achieves significant improvements in cross-domain adaptability and pseudo-label quality while delivering competitive recognition accuracy compared to state-of-the-art techniques. The source code is available at <uri>https://github.com/HawlsonZ/MEFANet</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14590-14603"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097383","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DPSFusion: Dynamic Pseudo-Supervision and Semantic Guidance for Multi-Modality Image Fusion 多模态图像融合的动态伪监督和语义引导

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-01 DOI: 10.1109/TCSVT.2026.3708884

Tiantian Yan;Shaozhe Zhang;Jiexiang Fang;Jinyuan Liu;Xin Yang;Qiang Zhang;Dongsheng Zhou

{"title":"DPSFusion: Dynamic Pseudo-Supervision and Semantic Guidance for Multi-Modality Image Fusion","authors":"Tiantian Yan;Shaozhe Zhang;Jiexiang Fang;Jinyuan Liu;Xin Yang;Qiang Zhang;Dongsheng Zhou","doi":"10.1109/TCSVT.2026.3708884","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3708884","url":null,"abstract":"Existing multimodal image fusion methods mainly emphasize network design and fusion rules to combine complementary information from different modalities. However, the supervision information adopted during training is often not aligned with the evaluation metrics employed to assess fused images. As a result, this misalignment biases optimization toward pixel similarity rather than perceptual quality and detail preservation, which causes blurred edges, texture loss, and modality imbalance. In addition, when task labels are unavailable, many approaches fail to fully exploit semantic relations between modalities. This weak alignment of cross-modal features leads to missing salient regions. To address these issues, we propose DPSFusion, a novel fusion framework that integrates a dynamic pseudo-supervision (DP) mechanism and a semantic guidance (SG) module. Specifically, the DP mechanism constructs pseudo-supervision images from diverse candidate fused results and adaptively identifies high-quality References using a multi-metric Fusion Quality Score (FQS). The pseudo-supervision set is subsequently iteratively updated during training, which ensures that the optimization objective gradually aligns with the evaluation metrics employed to assess fusion quality. In parallel, SG module is designed to exploit semantics without task-specific labels in order to produce fused images with sharper structures and higher target saliency. It extracts multi-level embeddings from infrared, visible, and pseudo-supervision images using a Contrastive Language-Image Pre-training (CLIP) model and converts them into feature modulation inside the network to highlight modality strengths and suppress redundancy. Finally, experiments on MSRS, RoadScene, TNO, and M<sup>3</sup>FD show state-of-the-art results on six standard fusion metrics and improved downstream performance in semantic segmentation and object detection, demonstrating the effectiveness of DPSFusion.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15244-15259"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097442","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Dreamer XL: Towards High-Resolution Text-to-3D Generation via Trajectory Score Matching 梦想家XL：通过轨迹评分匹配实现高分辨率文本到3d生成

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-09 DOI: 10.1109/TCSVT.2026.3701506

Xingyu Miao;Haoran Duan;Zizhou Ouyang;Runze Hu;Tejal Shah;Varun Ojha;Yang Long;Rajiv Ranjan

{"title":"Dreamer XL: Towards High-Resolution Text-to-3D Generation via Trajectory Score Matching","authors":"Xingyu Miao;Haoran Duan;Zizhou Ouyang;Runze Hu;Tejal Shah;Varun Ojha;Yang Long;Rajiv Ranjan","doi":"10.1109/TCSVT.2026.3701506","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3701506","url":null,"abstract":"In this work, we propose Trajectory Score Matching (TSM), a method designed to address the pseudo-ground truth inconsistency caused by accumulated error in Interval Score Matching (ISM) with Denoising Diffusion Implicit Models (DDIM) inversion process. Unlike ISM which adopts the inversion process of DDIM to calculate on a single path, our TSM method leverages the inversion process of DDIM to generate two paths from the same starting point for calculation. Since both paths start from the same starting point, TSM can reduce the accumulated error compared to ISM, thus alleviating the problem of pseudo-ground truth inconsistency. TSM enhances the stability and consistency of the model’s generated paths during the distillation process. We demonstrate this experimentally and further show that ISM is a special case of TSM. Furthermore, to optimize the current multi-stage optimization process from high-resolution text to 3D generation, we adopt Stable Diffusion XL for guidance. In response to the issues of abnormal replication and splitting caused by unstable gradients during the 3D Gaussian splatting process when using Stable Diffusion XL, we propose a pixel-by-pixel gradient clipping method. Extensive experiments show significant improvement over the baseline methods in visual quality and performance, demonstrating the effectiveness of TSM and gradient clipping in enhancing model stability and output fidelity.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14361-14372"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097635","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reversible Data Hiding in Encrypted Images Using Adaptive Block Compression 使用自适应块压缩在加密图像中隐藏可逆数据

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-16 DOI: 10.1109/TCSVT.2026.3704511

Zi-Long Liu;Chi-Man Pun;Ka-Cheng Choi;Kwok-Yan Lam

引用次数: 0