检索结果-内蒙古大学图书馆

Deep Residual Network with Multi-image Attention for Imputing Under Clouds in Satellite imagery 26

Deep Residual Network with Multi-Image Attention for Imputin...

26th International Conference on pattern recognition / 8th International Workshop on image Mining - Theory and Applications (IMTA)

作者： Yang, Xian Zhao, Yifan Vatsavai, Ranga Raju North Carolina State Univ Dept Comp Sci Raleigh NC 27695 USA BReLL Lirio Knoxville TN USA North Carolina State Univ CGA CSC Raleigh NC 27695 USA

ISBN: (数字)9781665490627

ISBN: (纸本)9781665490627

Earth observations from remote sensing imagery play an important role in many environmental applications ranging from natural resource (e.g., crops, forests) monitoring to man-made object (e.g., builds, factories) recognition. Most widely used optical remote sensing data however is often contaminated by clouds making it hard to identify the objects underneath. Fortunately, with the recent advances and increased operational satellites, the spatial and temporal density of image collections have significantly increased. In this paper, we present a novel deep learning-based imputation technique for inferring spectral values under the clouds using nearby cloud-free satellite image observations. The proposed deep learning architecture, extended contextual attention (ECA), exploits similar properties from the cloud-free areas to tackle clouds of different sizes occurring at arbitrary locations in the image. A contextual attention mechanism is incorporated to utilize the useful cloud-free information from multiple images. To maximize the imputation performance of the model on the cloudy patches instead of the entire image, a two-phase custom loss function is deployed to guide the model. To study the performance of our model, we trained our model on a benchmark Sentinel-2 dataset by superimposing real-world cloud patterns. Extensive experiments and comparisons against the state-of-the-art methods using pixel-wise and structural metrics show the improved performance of our model. Our experiments demonstrated that the ECA method is consistently better than all other methods, it is 28.4% better on MSE and 31.7% better on cloudy MSE as compared to the state-of-the-art EDSR network.

关键词： Earth Deep learning Cloud computing Satellites Clouds Benchmark testing Data models

来源：评论

学校读者我要写书评

暂无评论

remote sensing image Registration Via Feature Fusion and Enhanced Matching 6

Remote Sensing Image Registration Via Feature Fusion and Enh...

引用

6th International Conference on Electronic Information Technology and Computer Engineering, EITCE 2022

作者： Li, Xianjing Chen, Ying Zhang, Wencheng Wang, Jiahao Wang, Wei School of Computer Science and Information Engineering Shanghai Institute of Technology Shanghai China

ISBN: (纸本)9781450397148

image registration is a basic problem in image analysis and image processing. image registration has important applications in aerial image fusion, pattern recognition, three-dimensional reconstruction and other fields. Aiming at the problem of low registration accuracy and mismatching in remote sensing image registration, this paper proposes to use the Involution kernel to improve the ResNext network in the feature extraction stage and combines the SPANet attention mechanism with an improved ResNext network to improve the feature extraction ability of the network. In the feature matching stage, an enhanced matching method is proposed, which uses cross-correlation and nearest neighbor to second nearest neighbor ratio to filter out mismatched points to cope with complex images and background interference. The experimental results show that the proposed algorithm can achieve superior results in a variety of indexes compared with other algorithms, which proves that the proposed algorithm is effective. © 2022 Association for Computing Machinery.

关键词： remote sensing

来源：评论

学校读者我要写书评

暂无评论

Parallel Enhanced Network for image Compressed sensing

Parallel Enhanced Network for Image Compressed Sensing

引用

2022 International Conference on image, Signal processing, and pattern recognition, ISPP 2022

作者： Ma, Wenna Wu, Xiangjun College of Software Henan University Kaifeng475004 China

ISBN: (纸本)9781510654846

Despite the remarkable progress has made in deep compressed sensing (DCS), how to improve the reconstruction quality is still a major challenge. The existing DCS model generally still has some issues, especially in recovering details. In this paper, a new parallel enhanced network (PENet) is proposed for image compressed sensing. PENet is designed as a sampling network and a parallel network, which contains a basic network and an enhanced network. The basic network is designed to provide the initial reconstructed image. The enhanced network is trained to progressively acquire module details through the connections with each block of the basic network in stages. The final reconstructed image is the cumulative results between the parallel network. Experimental result shows that PENet has a high reconstruction quality and comparable running time complexity with existing advanced DCS methods. © 2022 SPIE.

关键词： Compressed sensing

来源：评论

学校读者我要写书评

暂无评论

基于伪全局Swin Transformer的遥感图像识别算法

引用

模式识别与人工智能 2023年第9期36卷 818-831页

作者：王科平左鑫浩杨艺费树岷河南理工大学电气工程与自动化学院焦作454003 河南理工大学河南省智能装备直驱技术与控制国际联合实验室焦作454003 东南大学自动化学院南京210096

如何在多目标并列的情况下,确定符合人类思维习惯的核心目标是遥感图像识别的关键之一.因此,在全局视野下,为各目标分配符合人类视觉习惯的注意力,是甄选核心目标的有效途径之一.文中结合Transformer提取全局特征的思想和Swin Transfor... 详细信息

如何在多目标并列的情况下,确定符合人类思维习惯的核心目标是遥感图像识别的关键之一.因此,在全局视野下,为各目标分配符合人类视觉习惯的注意力,是甄选核心目标的有效途径之一.文中结合Transformer提取全局特征的思想和Swin Transformer对图像栅格化处理可降低计算量的优点,提出基于伪全局Swin Transformer的遥感图像识别算法.构建伪全局Swin Transformer模块,将遥感图像栅格化后的各局部信息聚合为一个特征值,替代以像素为基础的全局信息,以较小计算量为代价,获取全局特征,有效提升模型对所有目标的感知能力.同时,通过以可变形卷积为基础的感受野自适应缩放模块,使感受野向核心目标偏移,提高网络对核心目标信息的关注,从而实现对遥感图像的精确识别.在RSSCN7、AID和OPTIMAL-31遥感图像数据集上的实验表明,文中算法取得较高的识别精度和参数识别效率.

关键词：遥感图像识别 Transformer Swin Transformer 核心目标

来源：评论

学校读者我要写书评

暂无评论

MRCI: Multi-range Context Interaction for Boundary Refinement in image Segmentation 27th

MRCI: Multi-range Context Interaction for Boundary Refineme...

引用

27th International Conference on pattern recognition, ICPR 2024

作者： Wu, Yaqiang Lyu, Wanjun Liang, Xianchen Zheng, Qinghua Wei, Jin Jin, Lianwen Xi’an Jiaotong University Xi’an China Lenovo Research Beijing China Beijing University of Posts and Telecommunications Beijing China South China University of Technology Guangzhou China

ISBN: (纸本)9783031801358

In the era of foundational image segmentation models, there is a pressing need to leverage the outputs of these models and enhance the boundary accuracy of domain-specific segmentation results using lightweight post-processing techniques. Numerous existing boundary refinement approaches neglect the significance of incorporating diverse contextual scopes and global knowledge, resulting in restricted adaptability to different coarse segmentation errors. Moreover, the prevailing models are often lacking in lightweight design. To address these challenges, we propose a novel framework called Multi-Range Context Interaction (MRCI) that aims to refine the boundaries of predicted masks by incorporating comprehensive context knowledge while maintaining computational efficiency. Our approach utilizes a multi-range context-aware strategy to extract more informative local features and incorporates global knowledge prompts to guide the boundary refinement process. Experimental results on the widely used Cityscapes, ADE20K and satellite remote sensing dataset SpaceNet demonstrate the effectiveness of our approach, achieving top-tier Average Precision (AP) and mean IoU among the current state-of-the-art boundary refinement models while utilizing only 4M parameters. The source code will be available. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Bragg-region recognition of high-frequency radar spectra based on deep learning and image fusion processing

引用

INTERNATIONAL JOURNAL OF remote sensing 2022年第18期43卷 6766-6782页

作者： Chuang, Laurence Zsu-Hsin Chen, Yu-Ru Chung, Yu-Jen Wu, Li-Chung Tien, Tsung-Mo Natl Cheng Kung Univ Inst Ocean Technol & Marine Affairs Tainan Taiwan Natl Chung Shan Inst Sci & Technol Elect Syst Res Div Taoyuan Taiwan Naval Acad Dept Marine Sci Kaohsiung Taiwan Natl Cheng Kung Univ Coastal Ocean Monitoring Ctr Tainan Taiwan Natl Kaohsiung Univ Sci & Technol Dept Marine Environm Engn Kaohsiung Taiwan Natl Chung Shan Inst Sci & Technol 481 6th Neighborhood Sec Jiaan Zhongzheng Rd Taoyuan 32546 Taiwan

In previous research, the identification of the Bragg region, which is often achieved using image recognition, was easily affected by ionospheric interference and different manual parameter settings. Strong disturbances from the ionosphere and other environmental noise may interfere with high-frequency radar (HFR) systems when determining first-order Bragg regions and thus may directly influence the surface current mapping performance. To avoid human intervention in first-order Bragg-region recognition, deep learning methods were used to extract multiple levels of feature abstractions of the first-order Bragg region. In this study, to assist in developing sufficient training data, a procedure integrating a morphological approach and Otsu's method was proposed to provide manual labelled data for a U-Net deep learning model. Additionally, several sets of activation functions and optimizers were used to achieve optimal deep learning model performance. The best combination of model parameters results in an accuracy of over 90%, an F1 score of over 80%, and an intersection over union (IoU) reaching 60%. The identification time of one image is approximately 70 ms. These results demonstrate that this deep learning model can predict the position of first-order Bragg regions under ionospheric interference and avoid being affected by strong noise that could cause prediction errors. Hence, deep learning and image fusion processing can effectively recognize the first-order Bragg regions under strong interference and noise and can thereby improve the surface current mapping accuracy.

关键词： Deep learning image processing high-frequency radar (HFR) Bragg-region recognition

来源：评论

学校读者我要写书评

暂无评论

26th International Conference on pattern recognition, ICPR 2022

26th International Conference on Pattern Recognition, ICPR 2...

引用

26th International Conference on pattern recognition, ICPR 2022

ISBN: (纸本)9783031377440

The proceedings contain 49 papers. The special focus in this conference is on pattern recognition. The topics include: Virtualization and 3D Visualization of Historical Costume Replicas: Accessibility, Inclusivity, Virtuality;datafication of an Ancient Greek City: Multi-sensorial remote sensing of Heloros (Sicily);geolocation of Cultural Heritage Using Multi-view Knowledge Graph Embedding;masonry Structure Analysis, Completion and Style Transfer Using a Deep Neural Network;PARTICUL: Part Identification with Confidence Measure Using Unsupervised Learning;Explaining Classifications to Non-experts: An XAI User Study of Post-Hoc Explanations for a Classifier When People Lack Expertise;motif-Guided Time Series Counterfactual Explanations;comparison of Attention Models and Post-hoc Explanation Methods for Embryo Stage Identification: A Case Study;graph-Based Analysis of Hierarchical Embedding Generated by Deep Neural Network;Explainability of image Semantic Segmentation Through SHAP Values;CIELab Color Measurement Through RGB-D images;comparing Feature Importance and Rule Extraction for Interpretability on Text Data;physics-Informed Neural Networks for Solar Wind Prediction;less Labels, More Modalities: A Self-Training Framework to Reuse Pretrained Networks;feature Transformation for Cross-domain Few-Shot remote sensing Scene Classification;a Novel Methodology for High Resolution Sea Ice Motion Estimation;Deep Learning-Based Sea Ice Lead Detection from WorldView and Sentinel SAR imagery;Uncertainty Analysis of Sea Ice and Open Water Classification on SAR imagery Using a Bayesian CNN;a Two-Stage Road Segmentation Approach for remote sensing images;YUTO Tree5000: A Large-Scale Airborne LiDAR Dataset for Single Tree Detection;spatial Layout Consistency for 3D Semantic Segmentation;reconstruction of Cultural Heritage 3D Models from Sparse Point Clouds Using Implicit Neural Representations;From TrashCan to UNO: Deriving an Underwater image Dataset to Get a More Consistent and

关键词：

来源：评论

学校读者我要写书评

暂无评论

26th International Conference on pattern recognition, ICPR 2022

26th International Conference on Pattern Recognition, ICPR 2...

引用

26th International Conference on pattern recognition, ICPR 2022

ISBN: (纸本)9783031377303

关键词：

来源：评论

学校读者我要写书评

暂无评论

Wildland Fire Detection and Monitoring using a Drone-collected RGB/IR image Dataset

Wildland Fire Detection and Monitoring using a Drone-collect...

引用

IEEE Applied imagery pattern recognition Workshop (AIPR)

作者： Chen, Xiwen Hopkins, Bryce Wang, Hao O'Neill, Leo Afghah, Fatemeh Razi, Abolfazl Fule, Peter Coen, Janice Rowell, Eric Watts, Adam Clemson Univ Sch Comp Clemson SC 29634 USA Clemson Univ Dept Elect & Comp Engn Clemson SC 29634 USA No Arizona Univ Sch Forestry Flagstaff AZ 84001 USA Natl Ctr Atmospher Res Boulder CO 80301 USA Desert Res Inst Reno NV 89512 USA US Forest Serv Pacific Wildland Fire Sci Lab Seattle WA 98103 USA

ISBN: (纸本)9781665477291

Drone-based Unmanned Aerial Systems (UAS) provide an efficient means for early detection and monitoring of remote wildland fires due to their rapid deployment, low flight altitudes, high 3D maneuverability, and ever-expanding sensor capabilities. Recent sensor advancements have made side-by-side RGB/IR sensing feasible for UASs. The aggregation of optical and thermal images enables robust environmental observation, as the thermal feed provides information that would otherwise be obscured in a purely RGB setup, effectively "seeing through" thick smoke and tree occlusion. In this work, we present Fire detection and modeling: Aerial Multi-spectral image dataset (FLAME 2) [1], the first ever labeled collection of UAS-collected side-by-side RGB/IR aerial imagery of prescribed burns. Using FLAME 2, we then present two image-processing methodologies with Multi-modal Learning on our new dataset: (1) Deep Learning (DL)-based benchmarks for detecting fire and smoke frames with Transfer Learning and Feature Fusion. (2) an exemplary image-processing system cascaded in the DL-based classifier to perform fire localization. We show these two techniques achieve reasonable gains than either single-domain video inputs or training models from scratch in the fire detection task.

关键词： Smoke

来源：评论

学校读者我要写书评

暂无评论

Incomplete multi-view subspace clustering based on robust matrix completion

引用

NEUROCOMPUTING 2025年 621卷

作者： Xing, Lei Zheng, Xinhu Lu, Yao Chen, Badong Xi An Jiao Tong Univ Natl Engn Res Ctr Visual Informat & Applicat Natl Key Lab Human Machine Hybrid Augmented Intell Xian Peoples R China Xi An Jiao Tong Univ Inst Artificial Intelligence & Robot Xian Peoples R China Hong Kong Univ Sci & Technol GZ Intelligent Transportat Thrust Syst Hub Guangzhou Peoples R China Beijing Inst Remote Sensing Beijing 100011 Peoples R China

Multi-view clustering (MVC) is a popular approach used in data mining and pattern recognition to enhance clustering performance by leveraging complementary information from different data sources or features. However, real-world multi-view data often suffer from incomplete views, where some samples lack representation on one or more views due to various faults during data collection or processing. Efficiently exploiting the connections between multiple views from such incomplete data presents a significant challenge compared to complete-view clustering. Although several methods have been proposed in recent years to address the problem of incomplete multi-view clustering (IMVC), existing approaches have not fully considered: (1) the differential negative effects associated with varying missing data rates across different views, and (2) the impact of intrinsic noise arising from the diverse quality of these views. To address this problem, we propose a novel subspace clustering framework based on robust matrix completion for incomplete and unlabeled multi-view data. The proposed model consists of two modules. The first is a regularized missing data completion module from an information theoretic perspective, and the second is a centralized subspace model, combining low rank properties. In addition, we develop an efficient iterative algorithm using half-quadratic (HQ) and linearized alternating direction method (ADM). Finally, we conduct experiments on five benchmark multi-view datasets to demonstrate that our proposed method outperforms state-of-the-art methods inmost datasets and metrics.

关键词： Correntropy Multi-view clustering Robust Learning Half-quadratic

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：