遥感影像云检测网络泛化性能研究：以DeepLabv3+为例     

彭龙康  刘励聪  陈学泓  陈晋  曹鑫  邱岳安 《遥感学报》2021,25(5):1169-1186

近年来,深度学习算法得到了长足的发展,并开始应用于云检测。但是深度神经网络模型参数众多,依赖大量训练样本,因此理解其泛化性能对于深度学习在不同遥感影像的实际应用具有重要的参考价值。本文以深度语义分割算法DeepLabv3+为例,以一组广泛使用的云标记数据集"L8 Biome"为验证数据,探讨该算法用于云检测时在不同的地表景观、空间分辨率和光谱波段组合的遥感影像上的泛化性能。云标记数据集"L8Biome"包含96景具有全球代表性的Landsat 8 OLI影像及相应的人工云掩膜,被广泛用于测试云和云阴影检测算法性能。首先,利用Landsat 8 OLI云标记数据集"L8 Biome",构建不同类型景观、不同空间分辨率、不同波段组合的训练影像集和测试影像集;其次基于不同训练样本集和测试集,评估了DeepLabv3+算法在不同情况下的云检测精度,并与Fmask算法作对比分析。研究结果表明:(1)使用全混合景观类型的训练集训练出来的云检测网络在总体检测精度(92.81%)与稳定度(标准差12.08%)上都优于使用单一景观类型的训练集训练得到的云检测网络,也优于Fmask的总体精度(88.75%)与稳定度(标准差17.34%),说明在构建深度学习算法的训练集时,应该尽可能包含多类型的地表景观;(2)将全混合景观训练集中剔除一类景观的样本(冰/雪景观除外)构建的"混合-1"训练集与全混合景观训练集训练的DeepLabv3+网络的云检测精度也相差不大,说明现有训练样本集已具备较强的景观泛化能力;(3)基于30 m空间分辨率的全混合景观训练样本集训练得到的DeepLabv3+云检测网络在不同分辨率(30 m、60 m、120 m、240 m)的测试集上云检测精度差异不大,都取得较好的效果,说明DeepLabv3+能够泛化应用于不同空间分辨率的遥感影像,相反Fmask直接应用于低分辨率影像时精度明显下降;(4) DeepLabv3+能充分自适应不同波段的信息用于云检测,总体来说更多的光谱波段输入能够提高DeepLabv3+的云检测的精度和稳定度,其中短波红外波段对于DeepLabv3+区分冰/雪与云具有重要价值,而热红外波段对DeepLabv3+云检测网络的性能提升很微小。以上结果说明利用现有数据集"L8 Biome"训练的DeepLabv3+云检测网络能够适用于多种类型的遥感影像,并优于Fmask算法。  相似文献   

机器学习在地下水环境背景值与污染风险评价的应用和展望     

曹文庚  付宇  南天  任宇  张栋  孙晓悦  翟文华  帅官印 《地质学报》2023,97(7):2408-2424

地下水资源在世界各国水资源中占有举足轻重的地位,对人类生存发展、维系生态系统健康发挥着重要作用。现阶段地下水污染日益严重,地下水环境背景值研究和污染风险评价对地下水污染防治工作具有重要意义。由于地下水污染影响因素复杂,地下水化学组分空间分布的非均质性、地下水样品采集的小样本问题与大尺度区域的高计算代价,都对传统的污染风险评价方法构成了极大挑战。机器学习作为人工智能的核心,已成为水文地质领域研究的前沿热点,通过智能高效的数据处理和挖掘,在地下水化学组分的分布、变化以及赋存机制等方向已得到探索和尝试。本文全面介绍了近年来在地下水污染研究方面应用的机器学习方法,涵盖了以聚类为主的非监督学习算法,以回归为主的监督学习算法,以提升算法效率为目标的混合算法,以及以神经网络为核心的深度结构算法,展示了不同类型算法在地下水污染研究方面的成果,详细归纳了各种算法的机理,对算法的技术优劣及适用方向进行了探讨;最后对机器学习在地下水污染方面的应用发展趋势进行了展望,建议探索高效集成学习模型,以弥补单一算法的不足,同时发展面向小样本的深度学习建模技术,提高地下水污染评价精度,拓展和丰富新方法新技术在地下水污染研究方面的应用。  相似文献   

高光谱图像相对含水量反演引导的精细分类     

梁雪剑  张晔  张钧萍 《遥感学报》2021,25(11):2283-2302

深度学习在高光谱图像处理领域的研究应用不断深入发展,基于深度学习的高光谱图像分类达到了较高的分类精度。目前的分类模型多利用高光谱的图谱特征,但对光谱的诊断性特征及先验信息利用不足,对空谱特征提取过程难以实现有效协同,因而导致分类类别即类内分类不够精细。为了解决以上问题,本文提出一种以多标签数据为输入的共生神经网络模型,在高光谱图谱特征提取的基础上融合光谱诊断特征,实现相对含水量反演及精细分类。首先,通过构建一种新的红边斜率光谱指数实现高光谱图像相对含水量的表征,利用本文提出的自适应分级算法完成相对含水量反演并建立对应的等级标签,与地物种类标签共同构成多标签高光谱数据集。然后,构建共生神经网络架构及内部变维特征提取模块,利用多标签数据提取高光谱图像中空间、光谱和相对含水量的融合特征,提高深度模型对不同含水量地物的区分能力和对所提取特征的协同表达能力,降低模型的复杂度与计算量,完成基于多标签数据集的相对含水量反演引导分类的过程,在扩大传统类间距离的基础上进一步扩大类内距离,从而实现高光谱图像的精细分类。最后,使用实验室采集数据和4个公开的高光谱数据集Lopex、Indian Pines、Pavia University和Salinas进行实验验证。结果表明,本文提出的红边斜率光谱指数可以有效表征地物的相对含水量信息;相对含水量反演引导的分类模型对类内分类精度有较明显的提升,对总体分类结果有一定的改善;与其他机器学习和深度学习分类算法相比,本文算法取得了较好的分类结果,提高了深度分类模型的分类性能和精细程度,实现了精细分类。  相似文献   

基于BDS短报文服务的海流机-岸基通信          下载免费PDF全文

王永鼎  陈小锋 《全球定位系统》2021,46(1):105-111

为了解决海流机海上作业的工况监测和海流机与岸基的远距离通信及数据传输困难等问题,选择将北斗卫星导航系统（BDS）应用到海流机在海上的定位和通信中. 分析海流特点和海流机的几种常见工况. 介绍基于北斗卫星导航系统-卫星无线电导航业务（BDS-RNSS）的海流机定位原理和BDS的短报文通信特点. 将“位拼接-LZW”双重数据压缩应用到短报文通信系统中. 设计一种适用于海流机通信的主控制系统并拟定海流机和岸上用户终端之间的通信方案,旨在提高通信效率和增大数据传输量.   相似文献   

三维点云孔洞修复方法综述     

赵江洪  孙铭悦  王殷瑞  窦新铜  张晓光 《测绘科学》2021,46(1):114-123

针对点云数据缺失问题,该文综合国内外大量点云修复技术研究。三维模型构建在自动驾驶、逆向工程领域中发挥越来越大的作用,三维点云数据是其中的重要数据源。利用三维激光扫描设备,可以高效、准确、实时的获取被测物体表面三维空间坐标。但是由于模型物体遮挡或者环境等原因,不可避免的会出现点云缺失的状况,这会对物体三维重建等后续处理造成一定的影响。然而,在三维点云孔洞修复方面还缺少比较系统完善的综述。本文从基于几何、基于模型检索、基于深度学习3个方面对当前主流的对修复技术进行了综合分析。文章对3种修复方法进行了概括,总结现有各种技术修复方法的优劣,同时展望了未来的发展趋势。  相似文献   

Value of incorporating geospatial information into the prediction of on-street parking occupancy – A case study     

Michael Balmer  Robert Weibel  Haosheng Huang 《地球空间信息科学学报》2021,24(3):438-457

In light of growing urban traffic, car parking becomes increasingly critical for cities to manage. As a result, the prediction of parking occupancy has sparked significant research interest in recent years. While many external data sources have been considered in the prediction models, the underlying geographic context has mostly been ignored. Thus, in order to study the contribution of geospatial information to parking occupancy prediction models, road network centrality, land use, and Point of Interest (POI) data were incorporated in Random Forest (RF) and Artificial Neural Network (ANN, specifically Feedforward Neural Network FFNN) prediction models in this work. Model performances were compared to a baseline, which only considers historical and temporal input data. Moreover, the influence of the amount of training data, the prediction horizon, and the spatial variation of the prediction were explored. The results show that the inclusion of geospatial information led to a performance improvement of up to 25％ compared to the baseline. Besides, as the prediction horizon expanded, predictions became less reliable, while the relevance of geospatial data increased. In general, land use and POI data proved to be more beneficial than road network centrality. The amount of training data did not have a significant influence on the performance of the RF model. The ANN model, conversely, achieved optimal results on a training input of 5 days. Likely attributable to varying occupancy patterns, prediction performance disparities could be identified for different parking districts and street segments. Generally, the RF model outperformed the ANN model on all predictions.  相似文献   

基于深度学习的立体影像密集匹配方法综述     

季顺平  罗冲  刘瑾 《武汉大学学报(信息科学版)》2021,(2)

三维重建可用于数字高程模型制作、机器人导航、增强现实和自动驾驶等。视差图是三维重建中一种重要的表达方式,而立体密集匹配是使用最广泛的获取视差图的技术。近年来,随着硬件、数据集、算法的发展,基于深度学习的立体匹配方法受到了广泛关注并取得了巨大成功。然而,这些方法通常在近景立体像对中进行测试,很少被用于遥感影像中。回顾了双目立体匹配的深度学习方法,选出了代表性的5种经典深度学习模型——GC-Net(geometry and context network)模型、PSM-Net(pyramid stereo matching network)模型、GWC-Net(group-wise correlation stereo network)模型、GA-Net(guided aggregation network)模型、HSM-Net(hierarchical deep stereo matching network)模型,将其应用于一套开源街景数据集(KITTI2015)和两套航空遥感影像数据集(München、WHU);分析了各种网络的实现方法,探讨了深度学习在遥感影像立体匹配中的性能,并与传统方法进行了对比。  相似文献   

极限学习机在黏土矿物分析中的应用     

白烨  曹乃文  邱庆良 《地质与资源》2021,30(4):505-511

对鄂尔多斯盆地苏里格北部上古生界盒8段低渗透储层研究发现,成岩作用是控制气藏分布的主要因素之一,而黏土矿物又影响着成岩作用类型及强度,在成岩作用划分中具有重要指示作用.本研究尝试利用自然伽马能谱测井结合神经网络（极限学习机）准确计算储层中黏土矿物含量,为全井自动成岩作用识别提供支撑.在黏土矿物分析过程中,为避免测井信息受岩石骨架颗粒成分差异的影响,针对研究区沉积的岩屑砂岩和岩屑石英砂岩分别建立黏土矿物分析神经网络,提高计算精度.神经网络采用了参数不易陷入局部最优的极限学习机,保证了分析结果的速度和稳定性.在此基础上,利用苏里格北部地区上古生界盒8段15个X衍射分析样本,将分岩性计算结果与及未分岩性分析结果进行比较,证明了方法的有效性.  相似文献   

Evaluation of deep learning algorithms for national scale landslide susceptibility mapping of Iran     

Phuong Thao Thi Ngo  Mahdi Panahi  Khabat Khosravi  Omid Ghorbanzadeh  Narges Kariminejad  Artemi Cerda  Saro Lee 《地学前缘(英文版)》2021,12(2):505-519

The identification of landslide-prone areas is an essential step in landslide hazard assessment and mitigation of landslide-related losses.In this study,we applied two novel deep learning algorithms,the recurrent neural network(RNN)and convolutional neural network(CNN),for national-scale landslide susceptibility mapping of Iran.We prepared a dataset comprising 4069 historical landslide locations and 11 conditioning factors(altitude,slope degree,profile curvature,distance to river,aspect,plan curvature,distance to road,distance to fault,rainfall,geology and land-sue)to construct a geospatial database and divided the data into the training and the testing dataset.We then developed RNN and CNN algorithms to generate landslide susceptibility maps of Iran using the training dataset.We calculated the receiver operating characteristic(ROC)curve and used the area under the curve(AUC)for the quantitative evaluation of the landslide susceptibility maps using the testing dataset.Better performance in both the training and testing phases was provided by the RNN algorithm(AUC=0.88)than by the CNN algorithm(AUC=0.85).Finally,we calculated areas of susceptibility for each province and found that 6%and 14%of the land area of Iran is very highly and highly susceptible to future landslide events,respectively,with the highest susceptibility in Chaharmahal and Bakhtiari Province(33.8%).About 31%of cities of Iran are located in areas with high and very high landslide susceptibility.The results of the present study will be useful for the development of landslide hazard mitigation strategies.  相似文献   

Probabilistic outlier detection for sparse multivariate geotechnical site investigation data using Bayesian learning     

Shuo Zheng  Yu-Xin Zhu  Dian-Qing Li  Zi-Jun Cao  Qin-Xuan Deng  Kok-Kwang Phoon 《地学前缘(英文版)》2021,12(1):425-439

Various uncertainties arising during acquisition process of geoscience data may result in anomalous data instances(i.e.,outliers)that do not conform with the expected pattern of regular data instances.With sparse multivariate data obtained from geotechnical site investigation,it is impossible to identify outliers with certainty due to the distortion of statistics of geotechnical parameters caused by outliers and their associated statistical uncertainty resulted from data sparsity.This paper develops a probabilistic outlier detection method for sparse multivariate data obtained from geotechnical site investigation.The proposed approach quantifies the outlying probability of each data instance based on Mahalanobis distance and determines outliers as those data instances with outlying probabilities greater than 0.5.It tackles the distortion issue of statistics estimated from the dataset with outliers by a re-sampling technique and accounts,rationally,for the statistical uncertainty by Bayesian machine learning.Moreover,the proposed approach also suggests an exclusive method to determine outlying components of each outlier.The proposed approach is illustrated and verified using simulated and real-life dataset.It showed that the proposed approach properly identifies outliers among sparse multivariate data and their corresponding outlying components in a probabilistic manner.It can significantly reduce the masking effect(i.e.,missing some actual outliers due to the distortion of statistics by the outliers and statistical uncertainty).It also found that outliers among sparse multivariate data instances affect significantly the construction of multivariate distribution of geotechnical parameters for uncertainty quantification.This emphasizes the necessity of data cleaning process(e.g.,outlier detection)for uncertainty quantification based on geoscience data.  相似文献