Hot spot tracking of f lood remote sensing research over the past 22 years:abibliometric analysis using CiteSpace

Background,aim,and scope In the context of climate change,extreme precipitation and resulting f looding events are becoming increasingly severe.Remote sensing technologies are advantageous for monitoring such disasters due to their wide observation range,periodic revisit capabilities,and continuous spatial coverage.These tools enable real-time and quantitative assessment of f lood inundation.Over the past 20 years,the field of remote sensing for f loods has seen significant advancements.Understanding the evolution of research hotspots within this field can offer valuable insights for future research directions.Materials and methods This study systematically analyzes the development and hotspot evolution in the field of f lood remote sensing,both domestically and internationally during 2000—2021.Data from CNKI(China National Knowledge Infrastructure)and WOS(Web of Science)databases are utilized for this analysis.Results(1)A total of 1693 articles have been published in this field,showing a stable growth trend post-2008.Significant contributors include the Chinese Academy of Sciences,Beijing Normal University,Wuhan University,the Italian National Research Council,and National Aeronautics and Space Administration.(2)High-frequency keywords from 2000 to 2021 include“remote sensing”“f lood”“model”“classification”“GIS”“climate change”“area”,and“MODIS”.(3)The most prominent keywords were“GIS”(8.65),“surface water”(7.16),“remote sensing”(7.07),“machine learning”(6.52),and“sentinel-2”(5.86).(4)Thirteen cluster labels were identified through clustering,divided into three phases:2000—2009(initial exploratory stage),2010—2014(period of rapid development),and 2015—2021(steady development of remote sensing for f loods and related disasters).Discussion The field exhibits strong phase-based development,with research focuses shifting over time.From 2000 to 2009,emphasis was on remote sensing image application and f lood model development.From 2010 to 2014,the focus shifted to accurate interpretation of remote sensing images,multispectral image applications,and long time series detection.From 2015 to 2021,research concentrated on steady development,leveraging large datasets and advanced data processing techniques,including improvements in water body indices,big data fusion,deep learning,and drone monitoring.Early on,SAR data,known for its all-weather capability,was crucial for rapid f lood hazard extraction and f lood hydrological models.With the rise of high-quality optical satellites,optical remote sensing has become more prevalent,though algorithm accuracy and efficiency for water body index methods still require improvement.Conclusions Data sources and methodologies have evolved from early reliance on radar data to the current exploration of optical image fusion and multi-source data integration.Algorithms now increasingly employ deep learning,super image elements,and object-oriented methods to enhance f lood identification accuracy.Recent studies focus on spatial and temporal changes in f looding,risk identification,and early warning for climate change-related f looding,including glacial melting and lake outbursts.Recommendations and perspectives To enhance monitoring accuracy and timeliness,UAV technology should be further utilized.Strengthening multi-source data fusion and assimilation is crucial,as is analyzing long-term f lood disaster sequences to better understand their mechanisms.

Effective distributed convolutional neural network architecture for remote sensing images target classification with a pre-training approach

How to recognize targets with similar appearances from remote sensing images(RSIs) effectively and efficiently has become a big challenge. Recently, convolutional neural network(CNN) is preferred in the target classification due to the powerful feature representation ability and better performance. However,the training and testing of CNN mainly rely on single machine.Single machine has its natural limitation and bottleneck in processing RSIs due to limited hardware resources and huge time consuming. Besides, overfitting is a challenge for the CNN model due to the unbalance between RSIs data and the model structure.When a model is complex or the training data is relatively small,overfitting occurs and leads to a poor predictive performance. To address these problems, a distributed CNN architecture for RSIs target classification is proposed, which dramatically increases the training speed of CNN and system scalability. It improves the storage ability and processing efficiency of RSIs. Furthermore,Bayesian regularization approach is utilized in order to initialize the weights of the CNN extractor, which increases the robustness and flexibility of the CNN model. It helps prevent the overfitting and avoid the local optima caused by limited RSI training images or the inappropriate CNN structure. In addition, considering the efficiency of the Na¨?ve Bayes classifier, a distributed Na¨?ve Bayes classifier is designed to reduce the training cost. Compared with other algorithms, the proposed system and method perform the best and increase the recognition accuracy. The results show that the distributed system framework and the proposed algorithms are suitable for RSIs target classification tasks.

A Multi Moving Target Recognition Algorithm Based on Remote Sensing Video

The Earth observation remote sensing images can display ground activities and status intuitively,which plays an important role in civil and military fields.However,the information obtained from the research only from the perspective of images is limited,so in this paper we conduct research from the perspective of video.At present,the main problems faced when using a computer to identify remote sensing images are:They are difficult to build a fixed regular model of the target due to their weak moving regularity.Additionally,the number of pixels occupied by the target is not enough for accurate detection.However,the number of moving targets is large at the same time.In this case,the main targets cannot be recognized completely.This paper studies from the perspective of Gestalt vision,transforms the problem ofmoving target detection into the problem of salient region probability,and forms a Saliency map algorithm to extract moving targets.On this basis,a convolutional neural network with global information is constructed to identify and label the target.And the experimental results show that the algorithm can extract moving targets and realize moving target recognition under many complex conditions such as target’s long-term stay and small-amplitude movement.

Efficient knowledge distillation for hybrid models:A vision transformer‐convolutional neural network to convolutional neural network approach for classifying remote sensing images

In various fields,knowledge distillation(KD)techniques that combine vision transformers(ViTs)and convolutional neural networks(CNNs)as a hybrid teacher have shown remarkable results in classification.However,in the realm of remote sensing images(RSIs),existing KD research studies are not only scarce but also lack competitiveness.This issue significantly impedes the deployment of the notable advantages of ViTs and CNNs.To tackle this,the authors introduce a novel hybrid‐model KD approach named HMKD‐Net,which comprises a CNN‐ViT ensemble teacher and a CNN student.Contrary to popular opinion,the authors posit that the sparsity in RSI data distribution limits the effectiveness and efficiency of hybrid‐model knowledge transfer.As a solution,a simple yet innovative method to handle variances during the KD phase is suggested,leading to substantial enhancements in the effectiveness and efficiency of hybrid knowledge transfer.The authors assessed the performance of HMKD‐Net on three RSI datasets.The findings indicate that HMKD‐Net significantly outperforms other cuttingedge methods while maintaining a significantly smaller size.Specifically,HMKD‐Net exceeds other KD‐based methods with a maximum accuracy improvement of 22.8%across various datasets.As ablation experiments indicated,HMKD‐Net has cut down on time expenses by about 80%in the KD process.This research study validates that the hybrid‐model KD technique can be more effective and efficient if the data distribution sparsity in RSIs is well handled.

Literature analysis of Dendrolimus disasters based on remote sensing monitoring

Due to global climate change,Dendrolimus pests and diseases seriously threaten the protec-tion of forestry plants and the safety of crops all over the world.This paper aims to discuss the research results and frontier progress of Dendrolimus disasters based on remote sensing monitoring,trying to find the occurrence characteristics of pests.In this paper,bibliometric methods and CiteSpace knowledge graphs were used to analyze the publication trend,highly cited documents,key research institutions,and high-frequency keywords of the extracted documents in the Web of Science(WOS)database.The following conclusions are drawn:(1)The amount of research in WOS is on the rise,but it has declined in recent years.The countries with strong influence in national cooperation are mainly the United States and China.(2)The United States Department of Agriculture-Agricultural Research Service(USDA ARS)and the Chinese Academy of Sciences have published a lot.This paper reviewed the research progress of high-frequency institutions.(3)The key research topics focus on remote sensing,agriculture,and environmental sciences.Besides,the research hotspots include remote sensing monitoring,climate change,spectral reflectance,vegetation index,and precision agriculture.Finally,we put forward the current challenges and development trends of remote sensing pest monitoring.This paper can provide a reference for the research on remote sensing monitoring of Dendrolimus disasters in the future.

基于改进YOLOv7的遥感图像小目标检测方法

针对遥感图像中小目标数量众多且背景复杂所导致的识别精度低的问题,提出了一种改进的遥感图像小目标检测方法。该方法基于改进的YOLOv7网络模型,将双级路由注意力机制加入至下采样阶段以构建针对小目标的特征提取模块MP-ATT(max pooling-attention),使得模型更加关注小目标的特征,提高小目标检测精度。为了加强对小目标的细节感知能力,使用DCNv3(deformable convolution network v3)替换骨干网络中的二维卷积,以此构建新的层聚合模块ELAN-D。为网络设计新的小目标检测层以获取更精细的特征信息,从而提升模型的鲁棒性。同时使用MPDIoU(minimum point distance based IoU)替换原模型中的CIoU来优化损失函数,以适应遥感图像的尺度变化。实验表明,所提出的方法在DOTA-v1.0数据集上取得了良好效果,准确率、召回率和平均准确率(mean average precision,mAP)相比原模型分别提升了0.4、4.0、2.3个百分点,证明了该方法能够有效提升遥感图像中小目标的检测效果。

基于改进YOLOv5的遥感图像目标检测

目前目标检测技术虽然已经趋于成熟,但是对遥感图像的检测仍存在不少挑战。针对遥感图像的背景复杂、目标尺度差异大、目标方向任意等特点造成目标检测精度低下的问题,提出一种基于改进YOLOv5的遥感图像目标检测算法。首先,构建一种联合注意力的多尺度特征增强网络,充分融合高低层特征,使特征层具有语义信息的同时包含丰富的细节信息,并在融合过程中利用设计的特征聚焦模块帮助模型选择关键特征,抑制无关信息。其次,使用感受野模块(RFB)对融合后的特征图进行更新,扩大特征图的感受野,减少特征信息损失。最后,对目标增加旋转角度,并采用圆形平滑标签将回归问题转化成分类问题,提高遥感目标定位的准确性。在用于航拍图像目标检测的大规模数据集(DOTA)上的实验结果表明,与YOLOv5算法相比,所提算法的交并比(Io U)为0.5和0.5~0.95时的平均精度均值(m AP@0.5和m AP@0.5∶0.95)分别提高了7.3和3.3个百分点,能够明显提高复杂背景下遥感图像目标的检测精度,并改善对遥感目标的漏检和误检情况。

基于卷积神经网络的遥感图像目标识别仿真

在遥感图像中,目标往往位于复杂的地物背景中,包括不同类型的植被、土地覆盖、建筑物等。上述复杂的地物背景对目标识别造成了困难。为了精准识别遥感图像目标,提出一种卷积神经网络下遥感图像目标识别算法。将暗通道原理和双边滤波算法有效结合,对遥感图像展开增强处理。统计分析遥感图像目标尺度范围,通过训练和测试卷积神经网络,得到最佳目标感兴趣区域尺度。确定目标感兴趣区域最佳尺度后,构建基于卷积神经网络的遥感图像目标识别架构,完成遥感图像目标识别。通过实验分析证明,采用所提算法可以有效提升遥感图像增强效果,具有较好的遥感图像目标识别性能。

改进YOLOv6的遥感图像目标检测算法

针对遥感图像背景复杂、目标普遍比较小且呈多尺度分布所导致的目标检测精度较低的问题,提出了一种改进YOLOv6的遥感图像目标检测算法。在骨干网络引入一种坐标注意力模块,以提高复杂背景下模型的特征提取能力和目标定位能力。提出一种上下文增强模块,使模型获取丰富的上下文信息,从而提升模型提取多尺度目标细节信息的能力。为了实现不同尺度特征的自适应融合,通过在颈网络引入一种自适应空间特征融合,提升了多尺度目标尤其是小目标的检测精度。将所提改进算法在遥感图像公开数据集DOTA-v1.0上进行训练并测试,实验结果表明,改进算法的收敛速度与收敛精度均优于原算法,其中AP值达到了54.6%,相比原算法提高了1.4个百分点,同时相比于一些其他目前先进的目标检测算法在精度和速度上均有提升,证明了改进算法的有效性。

改进的YOLOv5s遥感影像机场场面飞机小目标识别

遥感影像具有地物拍摄模糊以及背景环境复杂等特点,导致大面积地面物体识别准确率较低。针对此问题,提出了一种基于YOLOv5s改进网络模型。该模型对YOLOv5s中的骨干提取网络和颈部多尺度特征融合网络进行优化,引入Swin Transformer以获得更多关于目标物体的特征信息;同时对骨干网络中的模块做了修剪;此外,模型中还添加了坐标注意力机制来提升特征提取和融合效果。对于遥感数据集进行了小目标检测识别测试实验,改进后的YOLOv5s网络的mAP值为0.8375,比YOLOv5s网络模型提高了0.0225。实验结果表明,改进后的YOLOv5s网络模型对比YOLO系列网络和EfficientDet模型有效地提高了识别准确率、召回率以及mAP值,并且在训练时间上也比YOLOv5s减少了1/12。

改进YOLOv4的遥感图像目标检测算法

为有效解决遥感图像目标检测算法在复杂背景下的检测效果不佳的问题,提出一种改进YOLOv4的目标检测算法。设计一种跨阶段残差结构,替换原主干网络的简单残差结构,降低模型参数量和计算负担;引入CBAM注意力机制,加强CSP模块间有效特征交互;使用跨阶段分层卷积模块重构特征融合阶段对深层特征图的处理方式,防止网络退化和梯度消失;采用Mish激活函数,增强融合网络对非线性特征的提取能力。在RSOD、DIOR数据集上的实验结果表明,改进YOLOv4算法的测试mAP相比原YOLOv4算法分别高出4.5%、7.3%,其检测速度分别达到48 fps、45 fps,在保证实时性的同时检测精度有较大提升。

GCN引导模型视点的光学遥感道路提取网络

在光学遥感图像中,道路易受遮挡物、铺装材料以及周围环境等多重因素的影响,导致其特征模糊不清。然而,现有道路提取方法即使增强其特征感知能力,仍在特征模糊区域存在大量误判。为解决上述问题,本文提出基于GCN引导模型视点的道路提取网络(RGGVNet)。RGGVNet采用编解码结构,并设计基于GCN的视点引导模块(GVPG)在编解码器的连接处反复引导模型视点,从而增强对特征模糊区域的关注。GVPG利用GCN信息传播过程具有平均特征权重的特性,将特征图中不同区域道路显著性水平作为拉普拉斯矩阵,参与到GCN信息传播从而实现引导模型视点。同时,提出密集引导视点策略(DGVS),采用密集连接的方式将编码器、GVPG和解码器相互连接,确保有效引导模型视点的同时缓解优化困难。在解码阶段设计多分辨率特征融合(MRFF)模块,最小化不同尺度道路特征在特征融合和上采样过程中的信息偏移和损失。在两个公开遥感道路数据集中,本文方法IoU分别达到65.84%和69.36%,F1-score分别达到79.40%和81.90%。从定量和定性两方面实验结果可以看出,本文所提方法性能优于其他主流方法。

基于改进YOLOv8的遥感图像飞机目标检测研究

为解决遥感图像飞机目标检测时易出现检测精度低与漏检误检等问题,提出了一种基于YOLOv8算法的遥感图像飞机目标检测改进算法。首先,将坐标注意力机制模块嵌入卷积模块中,使其能提取复杂背景下的飞机小目标;然后,优化了检测头,去除了大的目标检测头,在提升小目标检测能力的同时减少算法的计算量;最后,使用WIoU作为改进的损失函数,以提高检测精度。实验表明,改进的YOLOv8算法能够有效提高飞机检测精度,可适用于遥感图像的飞机目标检测。

基于残差密集块的激光遥感图像中目标检测方法

为了提高对目标检测的效果,提出基于残差密集块的激光遥感图像中目标检测方法。首先,设计基于残差密集块的卷积神经网络,在设计ReLU激活函数并完成网络训练后,基于含噪激光遥感图像的初步特征提取结果,利用单个卷积展开卷积映射处理,抽取出潜在干净图像。然后,通过聚类处理的方式,得到激光遥感图像中车辆目标的显著图,再利用大律法,通过建立的特征比例关系的方式检测出其中的目标信息。实验结果表明,应用该方法有效滤除激光遥感图像中的噪声,并精准检测出激光遥感图像中的车辆目标。相比于3种传统方法,该方法检测结果均值误差的最小值仅为0.0156,说明该方法有效实现了设计预期。

特征协同与细粒度感知的遥感图像小目标检测

针对遥感图像中小目标多、排列密集导致的漏检问题,提出一种特征协同与细粒度感知的遥感图像小目标检测算法。首先,构造精细特征协同策略,通过智能调整卷积核参数,优化了特征间的交互和整合过程;通过精确控制信息流,实现从粗糙到精细的渐进式特征精化。在此基础上,本文设计一个细粒度感知模块,将感知注意力与移动反向卷积结合形成一个增强型检测头,显著增强网络对于极小尺寸物体的感知能力。最后,为了提升模型训练的效率,采用MPDIoU和NWD作为回归损失函数,解决位置偏差,加快模型收敛。在DOTA1.0数据集和DOTA1.5数据集上的实验结果表明,改进后算法相比于基准方法,平均精度分别提高7.4%和6.1%,相较于其他算法具有明显优势,显著改善遥感图像中小目标的漏检情况。

基于深度学习的遥感激光图像特征定位技术

为提高对激光遥感图像特征点定位的准确度,提出基于深度学习算法的遥感激光图像特征定位技术。采用高光谱图模型参数融合和光谱波段图像检测方法,对遥感激光图像进行多频段检测和稀疏化表示,提取多光谱和点云数据,进行不同信息维度的特征重组;采用深度学习算法,动态迭代搜索目标物理和几何特征点。采用图像分割技术聚类处理特征点的复合特征;利用无向图建模特征定位的邻域关系,根据特征聚类和邻域输出表达结果,实现对遥感图像的特征点准确定位。仿真结果表明,采用该方法进行遥感激光图像特征定位的分辨率水平较高,能实现反模糊化的目标特征点定位,定位准确率最高为0.92,点云噪声干扰下,偏移量最高为6∗10^(-3)。

基于ACE与YOLOv5的电力遥感图像检测算法

针对电力遥感图像采集时存在大量浓烟以及数据样本少等问题,提出一种基于对数变换的改进型自动色彩均衡与改进后YOLOv5s模型的遥感图像去雾检测算法,旨在通过提高数据集的图像质量进而提高检测网络的检测精度。构建的改进型自动色彩均衡对电力遥感图像去雾增强,并通过图像质量和特征提取两方面进行了实验数据对比,实验结果表明改进的自动色彩均衡算法优于其它算法。其次,通过YOLOv5s检测算法对增强后的数据集进行训练,引入mosaic数据增强算法,并通过构建ghost卷积模块和NAM注意力模块降低了网络参数、提升了网络检测精度。

基于cosSTR-YOLOv7的多尺度遥感小目标检测

针对地理空间遥感图像存在目标分布密集、尺度变化范围较大及小目标特征信息过少等而造成目标检测精度不高的问题,提出了一种基于Swin Transformer(STR)和YOLOv7的多尺度遥感小目标检测算法cosSTR-YOLOv7。以YOLOv7作为基线网络,首先,使用STR模块替换主干网络中的E-ELAN模块,并利用余弦注意力机制和后正则化方法将其改进为cosSTR模块,以提升模型训练的稳定性;其次,在Neck部分构建新的特征融合层,以减少特征信息丢失;然后,在预测部分增加小目标预测层,以提升模型对小目标的检测能力;最后,采用新的SIoU损失函数计算定位损失,以加快模型收敛速度。利用遥感数据集DIOR进行实验,实验结果表明,所提算法平均精度均值(mAP)达到92.63%,对比原YOLOv7算法提高了3.73个百分点,对多尺度小目标的检测性能有显著提高。

基于DFECANet的遥感图像飞机目标检测方法

针对现有的遥感图像目标检测方法中对小尺寸飞机目标的检测精度不高、特征信息传递不准确、信息交互不充分等问题,提出了一种基于可辨别特征提取和上下文感知的遥感图像飞机目标检测方法。设计了以可辨别特征提取模块为主体的主干网络,用以加强对多尺度飞机目标的特征提取;引入自适应特征增强模块,选择性关注小目标、优化特征信息的传递与信息交互;并设计了特征融合上采样模块对特征图进行上采样操作,用以提升高层语义信息的准确性。在DOTAv1数据集上的检测精度达到了95.2%,相较于YOLOv5s、SCRDet、ASSD等主流算法,飞机目标的检测精度提高了3.7%~18%。此外,该方法的检测速度以及模型参数量分别为147 fps和13.4 M,相较于当前主流算法具备较强的竞争力,满足在遥感背景下对飞机目标的实时检测需求。

基于样本增量学习的遥感影像分类

深度学习模型在遥感影像分类中取得了显著的成绩。随着新的遥感数据不断被采集,基于深度学习的遥感影像分类模型在训练新数据、学习新知识时,对旧数据的识别性能会下降,即旧知识遗忘。为帮助遥感影像分类模型巩固旧知识和学习新知识,提出一种基于样本增量学习的遥感影像分类模型——增量协同学习知识模型(ICLKM)。该模型由两个知识网络组成,第一个网络通过知识蒸馏保留旧模型的输出,缓解知识遗忘问题;第二个网络将新数据的输出作为第一个网络的学习目标,通过维护双网络模型的一致性有效地学习新知识。最后两个网络共同学习,通过知识协同策略生成更精确的模型。在两个遥感数据集NWPU-RESISC45和AID上的实验结果表明,相较于微调训练(FT)方法,ICLKM的准确率分别提升了3.53和6.70个百分点。可见ICLKM能够有效解决遥感影像分类的知识遗忘问题,不断提高对已知遥感影像的识别准确率。