检索结果-内蒙古大学图书馆

5th Chinese Conference on pattern recognition and Computer Vision (PRCV)

作者： Sun, Chengzhe Dui, Chun Wu, Jiangjiang Chen, Hao Natl Univ Def Technol Changsha Peoples R China Key Lab Nat Resources Monitoring & Nat Resources Changsha Peoples R China

ISBN: (纸本)9783031189159;9783031189166

Change detection is one of the main applications of remote sensing images. Pixel-to-pixel change detection using deep learning has been a hot research spot. However, the current approach are not effective enough to fuse deep semantic features and raw spatial information, and the network does not have the ability to perform long-distance information aggregation due to the limitation of the convolutional kernel size. In this manuscript, we propose a Siamese UNet with a dense attention mechanism, named SUDANet to do change detection for remote sensing images. SUDANet add a channel attention mechanism and a self-attention mechanism to the dense skip connection between encoder and decoder which enable the model to fuse feature information in channel dimensions and spatial dimensions. Graph attention module is also added at the end of the encoder, enabling the model to perform correlation analysis and long-distance aggregation of deep semantic features. The experimental results on LEVIR dataset show that our method outperforms the state-of-the-art change detection methods.

关键词： Change detection Deep learning Attention mechanism remote sensing (RS) images

来源：评论

学校读者我要写书评

暂无评论

MSAug: Multi-Strategy Augmentation for rare classes in semantic segmentation of remote sensing images

引用

DISPLAYS 2024年 84卷

作者： Gong, Zhi Duan, Lijuan Xiao, Fengjin Wang, Yuxi Beijing Univ Technol Beijing Key Lab Trusted Comp Beijing 100124 Peoples R China Beijing Univ Technol Natl Engn Lab Crit Technol Informat Secur Classifi Beijing 100124 Peoples R China Beijing Univ Technol Fac Informat Technol Beijing 100124 Peoples R China Beijing Climate Ctr Beijing 100081 Peoples R China Chinese Acad Sci Hong Kong Inst Sci & Innovat Ctr Artificial Intelligence & Robot Hong Kong 999077 Peoples R China Chinese Acad Sci Ctr Res Intelligent Percept & Comp Natl Lab Pattern Recognit Automat Beijing 100190 Peoples R China

Recently, remote sensing images have been widely used in many scenarios, gradually becoming the focus of social attention. Nevertheless, the limited annotation of scarce classes severely reduces segmentation performance. This phenomenon is more prominent in remote sensing image segmentation. Given this, we focus on image fusion and model feedback, proposing a multi-strategy method called MSAug to address the remote sensing imbalance problem. Firstly, we crop rare class images multiple times based on prior knowledge at the image patch level to provide more balanced samples. Secondly, we design an adaptive image enhancement module at the model feedback level to accurately classify rare classes at each stage and dynamically paste and mask different classes to further improve the model's recognition capabilities. The MSAug method is highly flexible and can be plug-and-play. Experimental results on remote sensing image segmentation datasets show that adding MSAug to any remote sensing image semantic segmentation network can bring varying degrees of performance improvement.

关键词： Data augmentation remote sensing images Semantic segmentation Rare classes

来源：评论

学校读者我要写书评

暂无评论

Point cloud restoration via 2D projection and inpainting 28

Point cloud restoration via 2D projection and inpainting

引用

Conference on image and Signal processing for remote sensing XXVIII

作者： Mai, Adrian Bilinski, Mark Provost, Raymond Hess, Michael NIWC Pacific 49275 Electron Dr San Diego CA 92152 USA

ISBN: (纸本)9781510655386;9781510655379

The presence of noise, displacement of points, and empty spots in a raw Light Detection and Ranging (LiDAR) point cloud are common phenomena caused by reflective surfaces or objects. Typical approaches to solve this problem are either avoid or cover the reflective areas or to manually remove the erroneous data in post processing. This can help clean the point cloud structure but will cause sparsity issues. To combat this, in this paper, we introduce a two-step process to perform point cloud restoration. Instead of removing noise, this approach can restore the points to the closest surface which they may belong to. Next, to fill out empty spots, we introduce a technique called point cloud inpainting, which involves interpolating points in 2D then mapping it back to 3D for flat surfaces. The point cloud then becomes more photorealistic and easier to use for other computer vision tasks.

关键词： Clouds 3D image processing image processing 3D acquisition Associative arrays Data processing LIDAR Machine vision Optical character recognition Scanners

来源：评论

学校读者我要写书评

暂无评论

Improved image recognition for AI Based Various Scientific Applications

Improved Image Recognition for AI Based Various Scientific A...

引用

2023 IEEE World AI IoT Congress, AIIoT 2023

作者： Karmakar, Supriyo Karmakar, Sayantani Farmingdale State College-SUNY FarmingdaleNY United States Portland State University PortlandOR United States

ISBN: (纸本)9798350337617

image processing has a lot of applications in our natural world. processing an image has applications in various fields like artificial intelligence (AI), Machine learning, remote sensing, image recovery and others. Various applications extract various information from images to successfully perform its operation and reach the solution. Overall, the quality of an image can be improved by increasing the signal-to-noise (SNR) of the image. This work shows the application of Savitzky-Golay (SG) Filter on a two-dimensional image increases the signal-to-noise (SNR) which is helpful to extract the required information for the next stage. This work also shows the applications of processed images in the AI based various scientific applications. © 2023 IEEE.

关键词： Decision trees

来源：评论

学校读者我要写书评

暂无评论

An Intelligent Registration Method of Heterogeneous remote sensing images Based on Style Transfer 5

An Intelligent Registration Method of Heterogeneous Remote S...

引用

5th International Conference on Artificial Intelligence and pattern recognition, AIPR 2022

作者： Tang, Haoyang Miao, Xin Shi, Jiakun Hua, Zhifan Yang, Dongfang School of Automation Xi'an University of Posts and Telecommunications Xi'an710121 China School of Computer Science Xi'an University of Posts and Telecommunications Xi'an710121 China The Rocket Force University of Engineering Graduate School Xi'an710025 China

ISBN: (纸本)9781450396899

Intelligent registration of heterogeneous remote sensing images is a hot issue in the field of remote sensing and has important research and application values. Due to the significant differences in data sources, image texture, color space and other factors of heterogeneous remote sensing images, the traditional image registration algorithms are difficult to be directly applied to the intelligent registration of images from different sources. In this paper, a new method of intelligent registration of heterogenous remote sensing images based on style transfer is proposed. First, the content features of the baseline image and the style features of the image to be aligned are extracted by using the generative adversarial nets, and the remote sensing images from different sources are fused in terms of content and style to obtain the remote sensing image transfer results with the same style as the image to be aligned and keeping the content of the baseline image unchanged. Then, an intelligent image registration algorithm is used to match the migrated remote sensing images. Finally, a heterogenous remote sensing image dataset is constructed around the heterogenous remote sensing image registration task, and the intelligent registration method of this paper is experimentally validated based on this dataset. The experimental results show that the intelligent image registration method based on style transfer can effectively improve the accuracy of heterogenous remote sensing image registration compared with the direct image registration method. © 2022 ACM.

关键词： image texture

来源：评论

学校读者我要写书评

暂无评论

Efficient recurrent attention network for remote sensing scene classification

引用

IET image processing 2021年第8期15卷 1712-1721页

作者： Liang, Le Wang, Guoli Data Commun Sci & Technol Res Inst Beijing Peoples R China Xingtang Telecommun Technol Co Ltd Beijing Peoples R China Tsinghua Univ Beijing 100000 Peoples R China

Scene classification for remote sensing is a popular topic, and many recent convolutional neural networks (CNNs)-based methods have shown the great model capacity and learning ability of highly discriminative features. Given a large number of training data, CNN can extract extensive features and learn to predict a remote sensing image. However, for supervised learning tasks, deep models often rely on a large number of labelled remote sensing images, which are difficult to pre-process. Thus, training a lightweight deep learning model is essential. Easy-classified and hard samples may also cause an imbalance of training set and lead the model to overwhelm the loss function. Accordingly, a novel Efficient Recurrent Attention Network (ERANet) for remote sensing scene classification is proposed. Different from traditional deep learning methods, Efficientnet-B0 is introduced as a lightweight backbone for the ARCNet framework, replacing the original one. By applying the modified efficient backbone, the low Floating Point Operations (FLOPs) and parameter numbers of the proposed ERANet are maintained. The significance of focal loss is determined and applied to address the sample imbalance problem and yield a desirable performance. Extensive experiments on several challenging remote sensing scene classification data sets prove the efficiency of the proposed ERANet.

关键词： Instrumentation and techniques for geophysical, hydrospheric and lower atmosphere research image recognition Geophysical techniques and equipment Computer vision and image processing techniques Geophysics computing Neural nets

来源：评论

学校读者我要写书评

暂无评论

Fusion of Hyperspectral and Multispectral images for Land Use Segmentation 9th

Fusion of Hyperspectral and Multispectral Images for Land Us...

引用

9th International Congress on Information and Communication Technology (ICICT)

作者： Irfan, Ayesha Beijing Univ Technol Fac Informat Technol Beijing Peoples R China

ISBN: (纸本)9789819735556;9789819735563

Land use feature detection is possible through high spatial and spectral resolution images. The spectral image resolution of hyperspectral sensors is mostly high, but the spatial resolution is low. The multispectral sensor produces good spatial resolution images with poor spectral resolution. In recent years, numerous techniques have been proposed for the improvement of spatial and spectral resolution in high-quality images acquired from the same scene by combining the acquired data. For this purpose, we fused hyperspectral and multispectral images, removed distortions, and improved the quality of the images by Gram-Schmidt, one of the pan-sharpening methods. Our selected study area is in Beijing, China. The fused image generated from the same area is utilized to segment land use through the U-Net model. The model we used for this study is trained and tested fused, multispectral, and hyperspectral images, and results showed that overall accuracy is 86.88 for hyperspectral imagery, 93.28 for multispectral imagery, and 97.59 for fused imagery. The results signify the employed approach has improved both image enhancement and fusion of images. Additionally, this technique can generate an imaging output characterized by high spectral data quality and high spatial resolution.

关键词： Land use segmentation pattern recognition remote sensing Hyperspectral Multispectral Data fusion

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Approaches for Crop Identification from remote sensing imagery: A Review 1

引用

14th International Conference on Soft Computing and pattern recognition, SoCPaR 2022, and the 14th World Congress on Nature and Biologically Inspired Computing, NaBIC 2022

作者： Wu, Falin Maleki, Reza Oubara, Amel Gómez, Dimaris Eftekhari, Akram Yang, Gongliu SNARS Laboratory School of Instrumentation and Optoelectronic Engineering Beihang University Beijing China School of Surveying and Geospatial Engineering College of Engineering University of Tehran Tehran Iran

ISBN: (数字)9783031275241

ISBN: (纸本)9783031275234

Crop area mapping as well as other remote sensing - based agricultural applications are vital for food security and production. This review investigates several studies published in two last years which focus on crop identification and classification using machine learning methods and remote sensing data. The popular conventional machine learning algorithms in crop classification are RF, SVM and LDA, and for deep learning algorithms are MLP, CNN and several different architectures of CNN include AlexNet, ResNet, Google-net, ResNet and U-net. RNN variations such as LSTM model are well-known deep learning architectures for processing time-series remote sensing data. Transfer learning builds a local classification algorithm to a new destination with a limited number of training samples. According to our investigation, the most widely used TL methods in recent researches are U-net (97% OA) and RF (90% OA) in agriculture scopes which have the best overall accuracies among the other methods. Also, more than 40 types of crop have been extracted in various researches, the main of which are corn, rice, wheat, soybeans, accounting for nearly 50% of all researches. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： remote sensing

来源：评论

学校读者我要写书评

暂无评论

Multi-Modal Ship Object Detection in Optical and SAR images 2

Multi-Modal Ship Object Detection in Optical and SAR Images

引用

2nd IEEE International Conference on image processing and Computer Applications, ICIPCA 2024

作者： Wang, Yong Li, Ling Tan, Xiao Ji, Bing Naval University of Engineering Department of Operational Research and Programming Wuhan China Engineering College of Field Engineering The Army Engineering University of Pla Nanjing China Naval University of Engineering Department of Navigation Wuhan China

ISBN: (数字)9798350360240

ISBN: (纸本)9798350360240

Ship object detection is one of the key technologies in the fields of maritime traffic monitoring, maintenance of maritime rights and interests, and coastal defense early warning detection. Optical remote sensing data contains rich texture information, which is conducive to target recognition, but it is highly depends on light and can only be used during the day. The influence of light on synthetic aperture radar (SAR) data is negligible, so it can be used all-weather, but it is difficult to accurately distinguish ship targets due to the lack of texture information. In order to comprehensively utilize two different data and fully leverage their respective advantages, a multimodal ship object rotated detection algorithm is proposed. Feature extraction is conducted separately for optical and SAR data, and an intermediate fusion strategy is adopted to achieve the fusion and detection of optical and SAR features. The introduction of CBAM (Convolutional Block Attention Module) in the backbone facilitates automatic identification of the weight of each feature channel during the learning process, which enhances the effect of feature extraction and suppresses noise and irrelevant information. The proposed multimodal detection algorithm has been validated on the constructed QXS-SAROPT-SHIP dataset. Compared with typical horizontal bounding box detection and single-modal detection algorithms, the accuracy rate of the proposed multimodal detection algorithm has been significantly improved. Specifically, the accuracy of the multimodal detection algorithm with CBAM module has been increased by 1.73%, demonstrating the effectiveness of the proposed method. © 2024 IEEE.

关键词： Optical remote sensing

来源：评论

学校读者我要写书评

暂无评论

SlotFusion: Object-Centric Audiovisual Feature Fusion with Slot Attention for remote sensing Scene recognition

SlotFusion: Object-Centric Audiovisual Feature Fusion with S...

引用

International Conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Fangzhou Han Tianyi Yu Lamei Zhang Lingyu Si Yiqi Zhang Dept. of Information Engineering Harbin Institute of Technology Harbin China National Key Laboratory of Space Integrated Information System Institute of Software Chinese Academy of Sciences Beijing China Institute of Software Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Despite significant advancements in remote sensing multimodal learning, particularly in image-image feature fusion, the exploration of audio-image feature fusion remains insufficient. Given the complexity and redundancy of ground objects in remote sensing images, accurately aligning audio features with image features during the fusion process is a critical challenge. In this paper, we introduce an object-centric feature fusion method named SlotFusion. By employing a slot attention-based feature decoupling module and a slot-based audiovisual feature fusion module, we transform modality features with complex semantic information into a set of slot features corresponding to object units and use gated activation units to adaptively implement object-centric feature fusion. Experiments on the Audio Visual Aerial Scene recognition dataset (ADVANCE) demonstrate that the proposed SlotFusion significantly improves remote sensing scene recognition performance, with a 7.04% increase in overall accuracy compared to previous methods, achieving state-of-the-art results.

关键词： Visualization Semantics Termination of employment Transforms Interference Signal processing Logic gates Robustness Speech processing remote sensing

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：