Preliminary analytical method for unsupervised remote sensing image classification based on visual perception and a force field

Abstract

The analysis of remote sensing (RS) images, which is often accomplished using unsupervised image classification techniques, requires an effective method to determine an appropriate number of classification clusters. This paper proposes a preliminary analytical method to evaluate the input parameters for unsupervised RS image classification. Our approach involves first analysing the colour spaces of RS images based on the human visual perception theory. This enables the initial number of clusters and their corresponding centres to be automatically established based on the interaction of different forces in our supposed force field. The proposed approach can automatically determine the appropriate initial number of clusters and their corresponding centres for unsupervised image classification. A comparison of the experimental results with those of existing methods showed that the proposed method can considerably facilitate unsupervised image classification for acquiring accurate results efficiently and effectively without any prior knowledge.     

Unmanned aerial systems for photogrammetry and remote sensing: A review

We discuss the evolution and state-of-the-art of the use of Unmanned Aerial Systems (UAS) in the field of Photogrammetry and Remote Sensing (PaRS). UAS, Remotely-Piloted Aerial Systems, Unmanned Aerial Vehicles or simply, drones are a hot topic comprising a diverse array of aspects including technology, privacy rights, safety and regulations, and even war and peace. Modern photogrammetry and remote sensing identified the potential of UAS-sourced imagery more than thirty years ago. In the last five years, these two sister disciplines have developed technology and methods that challenge the current aeronautical regulatory framework and their own traditional acquisition and processing methods. Navety and ingenuity have combined off-the-shelf, low-cost equipment with sophisticated computer vision, robotics and geomatic engineering. The results are cm-level resolution and accuracy products that can be generated even with cameras costing a few-hundred euros. In this review article, following a brief historic background and regulatory status analysis, we review the recent unmanned aircraft, sensing, navigation, orientation and general data processing developments for UAS photogrammetry and remote sensing with emphasis on the nano-micro-mini UAS segment.     

混合核特征加权SVM遥感湿地空间分类

针对湿地空间信息的复杂性和SVM的分类性能,设计一种基于混合核函数的特征加权SVM分类模型,综合利用多种特征信息,避免被弱相关特征所支配,从而提供更佳的映射性能和泛化能力。实验结果表明,该分类模型兼具良好的外推和内推能力,能够有效地融合不同信息源特征,得到更完整和准确的分类结果,在总体精度、Kappa系数等多项指标上都表现出更高的水平。     

A multi-level context-guided classification method with object-based convolutional neural network for land cover classification using very high resolution remote sensing images

Classification of very high resolution imagery (VHRI) is challenging due to the difficulty in mining complex spatial and spectral patterns from rich image details. Various object-based Convolutional Neural Networks (OCNN) for VHRI classification have been proposed to overcome the drawbacks of the redundant pixel-wise CNNs, owing to their low computational cost and fine contour-preserving. However, classification performance of OCNN is still limited by geometric distortions, insufficient feature representation, and lack of contextual guidance. In this paper, an innovative multi-level context-guided classification method with the OCNN (MLCG-OCNN) is proposed. A feature-fusing OCNN, including the object contour-preserving mask strategy with the supplement of object deformation coefficient, is developed for accurate object discrimination by learning simultaneously high-level features from independent spectral patterns, geometric characteristics, and object-level contextual information. Then pixel-level contextual guidance is used to further improve the per-object classification results. The MLCG-OCNN method is intentionally tested on two validated small image datasets with limited training samples, to assess the performance in applications of land cover classification where a trade-off between time-consumption of sample training and overall accuracy needs to be found, as it is very common in the practice. Compared with traditional benchmark methods including the patch-based per-pixel CNN (PBPP), the patch-based per-object CNN (PBPO), the pixel-wise CNN with object segmentation refinement (PO), semantic segmentation U-Net (U-NET), and DeepLabV3+(DLV3+), MLCG-OCNN method achieves remarkable classification performance (> 80 %). Compared with the state-of-the-art architecture DeepLabV3+, the MLCG-OCNN method demonstrates high computational efficiency for VHRI classification (4–5 times faster).     

Universal reconstruction method for radiometric quality improvement of remote sensing images

The performance of remote sensing images in some applications is often affected by the existence of noise, blurring, stripes and corrupted pixels, as well as the hardware limits of the sensor with respect to spatial resolution. This paper presents a universal reconstruction method that can be used to improve the image quality by performing image denoising, deconvolution, destriping, inpainting, interpolation and super-resolution reconstruction. The proposed method consists of two parts: a universal image observation model and a universal image reconstruction model. In the observation model, most degradation processes in remote sensing imaging are considered in order to relate the desired image to the observed images. For the reconstruction model, we use the maximum a posteriori (MAP) framework to set up the minimization energy equation. The likelihood probability density function (PDF) is constructed based on the image observation model, and a robust Huber–Markov model is employed as the prior PDF. Experimental results are presented to illustrate the effectiveness of the proposed method.     

面向对象的多尺度加权联合稀疏表示的高空间分辨率遥感影像分类

针对高空间分辨率遥感影像中的地物具有多尺度特性,以及各个尺度的对象特征对地物分类精度的影响具有较强的尺度效性,并结合面向对象影像分析方法和多尺度联合稀疏表示方法在高空间分辨率遥感影像分类中的各自优点,提出了一种面向对象的多尺度加权稀疏表示的高空间分辨率遥感影像分类算法。首先,采用多尺度分割算法获得多尺度分割结果并提取对象的多尺度特征;然后,根据影像对象的多尺度分割质量测度计算各尺度的对象权重,构建面向对象的多尺度加权联合稀疏表示模型;最后,采用2个国产GF-2高空间分辨率遥感数据集和1个高光谱-高空间分辨率航空遥感数据集（WashingtonD.C.数据）验证该算法的有效性。试验结果表明,与SVM、像素级稀疏表示、单尺度和多尺度对象级稀疏表示和深度学习等算法相比较,本文算法获得了较高的OA和Kappa分类精度,提高了各个尺度地物的分类精度,有效抑止了地物分类结果中的椒盐噪声现象,同时保持大尺度地物的区域性和小尺度地物的细节信息。     

高光谱遥感影像分类研究进展

随着模式识别、机器学习、遥感技术等相关学科领域的发展,高光谱遥感影像分类研究取得快速进展。本文系统总结和评述了当前高光谱遥感影像分类的相关研究进展,在总结分类策略的基础上,重点从以核方法为代表的新型分类器设计、特征挖掘、空间-光谱分类、基于主动学习和半监督学习的分类、基于稀疏表达的分类、多分类器集成六个方面对高光谱影像像素级分类最新研究进行了综述。针对今后的研究方向,指出高光谱遥感影像分类一方面要适应大数据、智能化高光谱对地观测的发展前沿,继续引入机器学习领域的新理论、新方法,综合利用多源遥感数据、多维特征空间互补的优势,提高分类精度、分类器泛化能力和自动化程度;另一方面要关注高光谱遥感应用的需求,突出高光谱遥感记录精细光谱特征的优势,针对应用需求发展有效的分类方法。     

JPEG 2000压缩对遥感影像监督分类精度的影响

遥感图像压缩会影响分类精度,是值得研究的问题。以高分辨率遥感影像(Quick Bird)的监督分类精度评定为尺度,采用ER Mapper软件的JPEG 2000图像压缩模块对图像进行压缩,再在eCognition软件中对这9种压缩比图像进行面向对象的监督分类,生成分类精度报告。通过分析分类精度的变化,研究了JPEG 2000压缩对遥感影像分类的影响程度及其在遥感影像压缩方面的应用潜力。     

区域多尺度马尔可夫随机场的遥感影像分类

多尺度分析技术广泛应用于高分辨率遥感影像的特征提取和建模。分解层数受制于影像的大小,下采样小波变换实现的影像多尺度表达难以描述大范围的空间模式,导致分类结果出现"胡椒盐"现象;面向对象的影像分析技术虽避免了"胡椒盐"现象,但由于仅利用了单尺度的的特征,也难以描述影像多层次的空间模式,导致分类精度较低。为改善分类结果中的"胡椒盐"现象和提高分类精度,提出了一种结合区域多尺度遥感影像分割和马尔可夫随机场的分类方法。首先,获得原始影像过分割区域,依据区域内亮度均值以及区域间的共享边界长度信息,提取影像低频和高频特征,采用该低频特征波段代替原始影像,重复分割与特征波段提取过程,形成影像的区域多尺度表达。然后,以原始图像为初始尺度,以分割区域为处理单元,以更细尺度分类结果为标记场先验,以当前高频特征建立特征场,逐层分类、投影,获得最终尺度分类结果。合成纹理影像和多光谱遥感影像的实验表明:相比于小波域多尺度建模方法和单尺度区域建模方法,本文提出的方法可以有效提高分类精度,并避免"胡椒盐"现象的产生。     

基于CNN-GCN双流网络的高分辨率遥感影像场景分类

高分辨率遥感影像具有复杂的几何结构和空间布局,传统的卷积神经网络的方法仅能提取场景图像中的全局特征,忽略了上下文的关系,导致特征的表达能力受限,制约了分类精度提高。针对此问题,本文提出一个面向高分辨率遥感影像场景分类的CNN-GCN双流网络,该算法包含CNN流和GCN流两个模块。CNN流基于预训练DenseNet-121网络提取高分影像的全局特征;而GCN流采用由预训练VGGNet-16网络得到的卷积特征图构建邻接图,再通过GCN模型提取高分影像的上下文特征。最后,通过加权级联的方式有效地融合全局特征和上下文特征并利用线性分类器实现分类。本文选取AID、RSSCN7和NWPU-RESISC45共3个具有挑战性的数据集进行实验,得到的最高分类精度分别是97.14%、95.46%和94.12%,结果表明本文算法能够有效地表征场景并取得具有竞争力的分类结果。     

A comprehensive review of earthquake-induced building damage detection with remote sensing techniques

Earthquakes are among the most catastrophic natural disasters to affect mankind. One of the critical problems after an earthquake is building damage assessment. The area, amount, rate, and type of the damage are essential information for rescue, humanitarian and reconstruction operations in the disaster area. Remote sensing techniques play an important role in obtaining building damage information because of their non-contact, low cost, wide field of view, and fast response capacities. Now that more and diverse types of remote sensing data become available, various methods are designed and reported for building damage assessment. This paper provides a comprehensive review of these methods in two categories: multi-temporal techniques that evaluate the changes between the pre- and post-event data and mono-temporal techniques that interpret only the post-event data. Both categories of methods are discussed and evaluated in detail in terms of the type of remote sensing data utilized, including optical, LiDAR and SAR data. Performances of the methods and future efforts are drawn from this extensive evaluation.     

Open land cover from OpenStreetMap and remote sensing

OpenStreetMap (OSM) tags were used to produce a global Open Land Cover (OLC) product with fractional data gaps available at osmlanduse.org. Data gaps in the global OLC map were filled for a case study in Heidelberg, Germany using free remote sensing data, which resulted in a land cover (LC) prototype with complete coverage in this area. Sixty tags in the OSM were used to allocate a Corine Land Cover (CLC) level 2 land use classification to 91.8% of the study area, and the remaining gaps were filled with remote sensing data. For this case study, complete are coverage OLC overall accuracy was estimated 87%, which performed better than the CLC product (81% overall accuracy) of 2012. Spatial thematic overlap for the two products was 84%. OLC was in large parts found to be more detailed than CLC, particularly when LC patterns were heterogeneous, and outperformed CLC in the classification of 12 of the 14 classes. Our OLC product represented data created in different periods; 53% of the area was 2011–2016, and 46% of the area was representative of 2016–2017.     

利用OpenStreetMap数据进行高空间分辨率遥感影像分类

针对高分辨率遥感影像分类样本标注困难的问题,提出了一种利用OpenStreetMap （OSM）数据自动获取标注样本的方法。与现有的利用OSM数据进行分类的方法不同,该方法加入了空间特征以弥补单独使用光谱特征分类的不足。首先,基于OSM数据提供的地物类别和位置信息进行样本标注,为了降低OSM数据中少量错误信息对分类结果的影响,采用聚类分析的方法对样本进行提纯;其次,使用形态学轮廓来提取影像的结构特征,挖掘高分辨率遥感影像丰富的空间信息,与光谱特征相叠加并输入分类器进行分类。试验证明,本文提出的方法能够有效避免人工样本标注所需要的人力物力;同时,联合影像的光谱空间特征能够更好地描述地物特性,得到较高的分类精度。     

A review of progress in identifying and characterizing biocrusts using proximal and remote sensing

Biocrusts are critical components of desert ecosystems, significantly modifying the surfaces they occupy. The mixture of biological components and soil particles that form the crust, in conjunction with moisture, determines the biocrusts’ spectral signatures. Proximal and remote sensing in complementary spectral regions, namely the reflective region, and the thermal region, have been used to study biocrusts in a non-destructive manner, in the laboratory, in the field, and from space. The objectives of this review paper are to present the spectral characteristics of biocrusts across the optical domain, and to discuss significant developments in the application of proximal and remote sensing for biocrust studies in the last few years. The motivation for using proximal and remote sensing in biocrust studies is discussed. Next, the application of reflectance spectroscopy to the study of biocrusts is presented followed by a review of the emergence of high spectral resolution thermal remote sensing, which facilitates the application of thermal spectroscopy for biocrust studies. Four specific topics at the forefront of proximal and remote sensing of biocrusts are discussed: (1) The use of remote sensing in determining the role of biocrusts in global biogeochemical cycles; (2) Monitoring the inceptive establishment of biocrusts; (3) Identifying and characterizing biocrusts using Longwave infrared spectroscopy; and (4) Diurnal emissivity dynamics of biocrusts in a sand dune environment. The paper concludes by identifying innovative technologies such as low altitude and high resolution imagery that are increasingly used in remote sensing science, and are expected to be used in future biocrusts studies.     

Support vector machines in remote sensing: A review

A wide range of methods for analysis of airborne- and satellite-derived imagery continues to be proposed and assessed. In this paper, we review remote sensing implementations of support vector machines (SVMs), a promising machine learning methodology. This review is timely due to the exponentially increasing number of works published in recent years. SVMs are particularly appealing in the remote sensing field due to their ability to generalize well even with limited training samples, a common limitation for remote sensing applications. However, they also suffer from parameter assignment issues that can significantly affect obtained results. A summary of empirical results is provided for various applications of over one hundred published works (as of April, 2010). It is our hope that this survey will provide guidelines for future applications of SVMs and possible areas of algorithm enhancement.     

Use of remote sensing to map gully erosion along the Atbara River, Sudan

The problems and impact of gully erosion along the Atbara River (Sudan), situated in semi-arid and arid environments, were investigated. The total gross area of gullied land and the loss of arable land by gully erosion were estimated. Multi-date sets of panchromatic aerial photographs and Landsat images (TM) were selected to represent two sites in the arid (New Halfa) and semi-arid (Showak town) zones along the Atbara River. Photo interpretation was conducted using physiographic and element methods. The interpretations detected the effects of water action in different climatic zones on geology, lithology, vegetation and land use. The results showed that the traditional rainfed agriculture has accelerated gully erosion in the semi-arid rather than in the arid zone. The progressive rate of gully erosion in the semi-arid zone resulted in loss of arable land at about 13.4 km² yr-1 and 9.8 km² yr-1 in the periods 1985–1987 and 1987–1990, respectively. The study provided data on the monitoring and mapping of gully erosion along the Atbara River and its tributaries.     

典型遥感数据分类方法的适用性分析——以遥感图像场景分类为例

目前普遍采用的分类器通常都是针对单一或小量任务而设计的,在小数据量的处理中能取得比较满意的结果。但对于海量遥感数据的处理,其在处理时效和分类精度方面还有待研究。本文以遥感图像场景分类任务为例,着重对遥感数据分类问题中几种典型分类方法的适用性进行比较研究,包括K近邻(KNN)、随机森林(RF),支持向量机(SVM)和稀疏表达分类器(SRC)等。分别从参数敏感性,训练样本数据量,待分类样本数据量和样本特征维数对分类器性能的影响等几个方面进行比较分析。实验结果表明:(1)KNN,RF和L0-SRC方法相比RBF-SVM,Linear-SVM和L1-SRC,受参数影响的程度更弱;(2)待分类样本固定的情况下,随着训练样本数目的增加,SRC类型分类方法的分类性能最佳,SVM类型方法次之,然后是RF和KNN,在总体分类时间上呈现出L0-SRCL1-SRCRFRBF-SVM/Linear-SVMKNN/L0-SRC-Batch的趋势;(3)训练样本固定的情况下,所有分类方法的分类精度几乎都不受待分类样本数目变化的影响,RBF-SVM方法性能最佳,其次是L1-SRC,然后是Linear-SVM,最后是RF和L0-SRC/L0-SRC-Batch,在总体分类时间上,L1-SRC和L0-SRC相比其他分类方法最为耗时;(4)样本特征维数的变化不仅影响分类器的运行效率,同时也影响其分类精度,其中SRC和KNN分类器器无需较高的特征维数即可获得较好的分类结果,SVM对高维特征具有较强的包容性和学习能力,RF分类器对特征维数增加则表现得并不敏感,特征维数的增加并不能对其分类精度的提升带来更多的贡献。总的来说,在大数据量的遥感数据分类任务中,现有分类方法具有良好的适用性,但是对于分类器的选择应当基于各自的特点和优势,结合实际应用的特点进行权衡和选择,选择参数敏感性较小,分类总体时间消耗低但分类精度相对较高的分类方法。     

油松毛虫灾害遥感监测及其影响因子分析

辽宁西部大面积的油松(Pinus tabulaeformis)人工林长期受到油松毛虫(Dendrolimus tabulaeformis)的危害,通过遥感技术,可以及时、高效、精准地对此大面积灾害进行监测,并获知地形、气象因子对其的影响。本文利用遥感和地理信息系统(GIS)技术,使用TM、ETM+数据,通过近红外与红光波段反射率的比值RVI,对油松的受灾程度进行了有效监测。前人的研究发现:油松毛虫易在干燥、温暖的环境爆发,本文将监测分类结果与地形、气象数据叠加后,分析发现结果亦与油松毛虫的生物学特性相吻合,由此逆向证明了监测结果的可靠性。通过对影像灰度直方图的分析,发现近红外波段对轻度的虫害敏感;红光波段对重度的虫害敏感。对影响因子的分析发现:油松毛虫在阳坡,坡度缓的地区危害更剧烈;在日照时数长、降雨少、积温低的地区,油松的受灾程度更重。此结论为预测虫害爆发的概率提供了依据。本研究表明:在森林灾害的遥感工作中,利用监测对象的生物生态学特性,可以在实地调查数据不足,难以直接对监测结果进行评价的情况下,判断结果的可靠性。利用此方法,一定程度上可以减少调查的工作量,降低外业的难度。     

多图卷积网络的遥感图像小样本分类

小样本学习旨在利用非常少的监督信息识别出新的类别,由于忽视了样本之间的关联信息,现有的小样本分类方法用于遥感图像小样本分类时往往不能获得令人满意的精度。为此,本文利用图来建模图像在特征空间的相似关系,使用图卷积运算平滑同类别图像的特征,增强不同类别图像特征的区分度,提升分类精度。所提方法在现有图卷积运算的基础上,使用多阶次的邻接矩阵线性加权的方法代替传统的一阶邻接矩阵,通过图谱分析得出这种改进方法能够让不同阶次邻接矩阵的频率响应函数在高频部分正负相抵,有效抑制图信号的高频分量,更显著的提升同类别节点特征的聚集程度;同时,在训练过程引入了微调的方法,使用新类别中的标记数据对最后一层图卷积网络进行少量次数的训练,能够进一步提高精度,增强模型的迁移能力。实验使用AID、OPTIMAL31以及RSI-CB256这3个常用的遥感数据集对方法的有效性进行了测试,结果表明提出的方法在同数据集小样本分类任务和跨数据集小样本分类任务中,在分类精度方面均优于原型网络等比较方法。     

Spectrodirectional remote sensing: From pixels to processes

This paper discusses the historical evolution of imaging spectroscopy in Earth observation as well as directional (or multiangular) research leading to current achievements in spectrodirectional remote sensing. It elaborates on the evolution from two separate research areas into a common approach to quantify the interaction of light with the Earth surface. The contribution of spectrodirectional remote sensing towards an improved understanding of the Earth System is given by discussing the benefits of converging from individual pixel analysis to process models in the land-biosphere domain. The paper concludes with an outlook of research focus and upcoming areas of interest emphasizing towards multidisciplinary approaches using integrated system solutions based on remote and in situ sensing, data assimilation, and state space estimation algorithms.