基于变化检测的滑坡灾害自动识别

在分析当前滑坡灾害信息识别方法的基础上,指出这些方法的适用性和不足之处,并针对这些不足之处,以滑坡的地学原理为依据,提出了以多时相遥感影像为数据源,结合纹理分析的变化检测自动识别滑坡灾害信息的方法。最后,以地震前(2006-5-14)后(2008-6-3)的北川县城及附近区域的福卫2号多光谱遥感影像作为数据源,以结合比率变换和纹理分析的变化检测进行滑坡灾害信息识别试验。结果表明,结合纹理分析的变化检测方法能够突破影像光谱特征的局限,对于滑坡灾害信息识别具有显著的适用性。     

基于形态学开运算的面向对象滑坡提取方法研究

山体滑坡是最常见的地质灾害之一,快速准确地获取滑坡的危害程度和分布情况对灾后救援具有重要意义。通过引入形态学开运算和区域水平集算法来构建面向对象的滑坡提取方法,以高分二号卫星的多光谱影像为数据源,在位于尼泊尔巴格马蒂专区的研究区中,分别运用构建的方法和基于像元的滑坡提取方法进行滑坡提取实验,并对两种方法的滑坡提取结果进行精度分析。实验结果表明:面向对象的滑坡提取方法总体上比基于像元的滑坡提取方法的提取结果精度更高,对云层和积雪的抗干扰能力更强。     

基于光谱、空间和形态特征的面向对象滑坡识别

地震经常诱发数量众多,覆盖范围广的滑坡并造成极大危害,因此,需要对大范围的滑坡灾害进行快速评估。随着遥感影像分辨率的提高,面向对象分类方法在这方面的应用比传统的目视解译和基于像素的方法更具优势。但是,目前基于面向对象方法的滑坡识别研究还相对较少,而且通常针对小范围的研究区。基于SPOT5 2.5m多光谱影像,提出一种综合光谱、空间、地形和形态特征的面向对象滑坡自动识别方法,并应用于较大范围研究区。结果表明:面向对象滑坡自动识别方法能将研究区内95%的滑坡识别出来,综合考虑滑坡的过度提取与遗漏提取情况,滑坡提取质量为74.04%,效果较好,能够快速、有效地识别大范围的滑坡。该方法可以应用于对地震或强降雨引起的大范围滑坡灾害进行快速评估,为灾后应急救援和恢复重建工作提供参考。     

林地山区滑坡遥感的最优融合方法及应用

滑坡作为造成巨大经济损失和人员伤亡的地质灾害之一,越来越引起社会的高度重视。为精确识别林地山区中的滑坡灾害,以2020年7月6日发生在湖南省常德市石门县南北镇潘坪村的雷家山滑坡为研究对象,使用不同的融合方法进行Sentinel-1A C波段干涉宽幅的地距多视产品和Sentinel-2A多光谱2A级光学影像融合,得到主成分分析融合方法对分贝化处理后的S1A VV极化影像与S2A影像融合效果最优,采用支持向量机方法分别对最优融合影像和原始S2A影像进行滑坡识别,最后使用S2A影像滑坡目视解译结果为检验标准对支持向量机滑坡识别结果进行精度评价,同时以2020年7月21日发生在湖北恩施屯堡乡马者村的沙子坝滑坡作为案例检验该方案的可推广性。结果表明：与单独使用光学影像进行研究区滑坡识别相比,使用最优融合影像滑坡识别的准确率由95.24%提升到了96.65%,滑坡提取质量也由87.18%提升到了91.84%,滑坡的漏识别和过度识别均有所降低,说明光学影像和合成孔径雷达影像融合的研究方案具有可推广性,能提高林地山区滑坡识别的准确率,可以更好地为滑坡风险评估、灾后应急调查以及灾后恢复重建等提供有价值的...     

基于多时相Landsat8 OLI影像的作物种植结构提取

针对基于多时相遥感影像、多种特征量提取多种作物种植结构在我国研究较少的现状,利用多时相Landsat8OLI影像数据,根据温宿县不同作物的农事历,通过分析主要地物的光谱特征和归一化植被指数的时间变化信息,构建不同作物种植结构提取的决策树模型,实现了对温宿县多种作物种植结构信息的提取。结果表明:1水稻的最佳识别依据是5月20日影像的近红外波段和7月23日影像的NDVI值;棉花和春玉米的最佳识别依据是5月20日~9月9日影像的NDVI变化值;冬小麦—夏玉米和林果的最佳识别依据是5月20日~7月23日影像的NDVI变化值;2与单时相监督分类相比,多时相决策树法对多种作物种植结构的提取效果更理想,总体精度提高了7.90%,Kappa系数提高了0.10;3Landsat8OLI影像数据分辨率高、成本低、获取方便,是农作物遥感的良好数据源。     

基于多源多时相遥感影像的山地森林分类决策树模型研究

山地是森林重要的分布区,然而山地多样的森林类型、高度异质化的景观格局、突出的地形效应以及云、雾的干扰均不同程度地影响了山地森林类型的遥感自动制图。多源多时相遥感影像提供的季相节律信息是当前提高土地覆被遥感制图精度的重要信息源之一。以岷江上游地区为研究区,以国产环境减灾卫星多光谱CCD(简称HJ CCD)影像和美国Landsat TM影像为数据源,以决策树为分类方法,根据参与分类影像的时相差异设计了5组对比实验(生长季单时相组、非生长季单时相组、生长季多时相组、非生长季多时相组、全时相组),对比论证多源多时相遥感影像对山地森林类型自动制图的贡献和作用。对比结果表明:生长季和非生长季相结合的多时相遥感影像较单时相或单一类型(生长季或非生长季)多时相遥感影像,更能显著提高山地森林类型自动制图精度,且能降低分类决策树的复杂程度,更有利于山地森林类型的自动提取。     

基于面向对象方法的沙化土地遥感信息提取技术研究

针对基于像元光谱特征提取沙化土地信息分类精度偏低的问题,以Landsat\|5 TM为数据源,基于面向对象的方法对沙化土地遥感信息提取技术进行研究。首先采用多尺度分割法对影像进行分割以获得同质区域,然后结合野外调查数据制成不同地物类型的多种特征图,从而确定提取目标地物的特征并建立沙化和非沙化地物提取决策树,最后对影像进行模糊分类,并对分类结果进行精度评价。结果表明,基于面向对象提取沙化土地信息的总精度达84.89%,Kappa系数为0.8077。研究结果为后续深入研究奠定了基础。     

基于PCA与EM算法的多光谱遥感影像变化检测研究

变化检测技术能够从观测数据中有效地提取各种变化信息。以多时相的多光谱遥感影像作为主要研究对象,就基于直接比较像元特征的变化检测中两方面的关键问题——差异影像的构造和变化阈值的提取来开展研究,具体采用PCA结合差值比较的方法构造差异影像,采用EM自动确定变化阈值以提取变化区域。通过与传统变化检测方法的实验比较,证明了该方法的有效性。     

面向对象变化向量分析的遥感影像变化检测

为了解决基于像元的变化向量分析法在高分辨率遥感影像变化检测中精度低的问题,提出了一种面向对象变化向量分析的遥感影像变化检测方法。综合2个时期的遥感影像,首先通过影像分割获取像斑,其次提取直方图作为像斑的特征向量,再次采用直方图相交法度量2个时期像斑直方图之间的距离,构建像斑的变化向量,然后利用加权组合的方法计算像斑变化向量的模,最后依据最大熵原理获取变化检测阈值,对像斑进行变化/未变化判别。在QuickBird及Ikonos遥感影像上的实验表明:在高分辨率遥感影像变化检测中,与基于像元的变化向量分析法相比,该方法变化检测的精度较优,变化检测的正确率分别达到了0.92与0.90。     

结合改进混合卷积模型的遥感影像变化检测

针对基于深度学习的变化检测模型搭建中提高变化检测精度这一难点,在综合考虑面向像元和面向对象变化检测算法的基础上,设计了一种基于改进混合卷积特征提取模块的变化检测模型。该模型结合多切片思想和并行神经网络结构,融合不同尺寸的卷积核获取丰富的多尺度特征。首先,利用超像素分割算法将测试影像分割成无重叠的同质性区域;然后,选取一定数量的样本对模型进行训练,得到测试影像的像素级变化检测结果;最后,利用投票法,将网络得到的像素级结果与分割对象相结合,得到最终的变化检测结果。实验结果表明,基于该方法的网络模型性能较好,该模型可以有效学习多时相影像中的空间信息及差异特征,同时结合分割算法能够降低虚检率和漏检率,有效提高了变化检测精度。     

基于多特征面向对象区域滑坡现象识别

高空间分辨率卫星遥感数据的发展为滑坡灾害数据获取和更新提供了新的途径。以西北黄土高原为研究区,提出了一种基于多特征面向对象区域滑坡现象的识别方法,基于单期高空间分辨率遥感数据,利用集合和特征组合进行区域滑坡现象识别,实验结果表明:该方法是识别滑坡现象有效的技术方法之一,对开展滑坡监测、影像理解和地学分析具有重要的研究意义。     

基于面向对象的多源卫星遥感影像玉米倒伏面积提取

风灾引起的玉米倒伏可能导致玉米大量减产,利用遥感技术准确监测玉米倒伏面积与空间分布信息对灾情的评估非常重要。利用Planet和Sentinel-2影像分别结合面向对象与基于像元方法提取研究区玉米倒伏,同时评估了不同影像特征（光谱特征、植被指数和纹理特征）与不同分类方法（支持向量机法SVM、随机森林法RF和最大似然法MLC）对玉米倒伏提取精度的影响。结果表明：(1)使用高空间分辨率的Planet影像进行玉米倒伏提取的精度普遍高于Sentinel-2影像;(2)从分类精度和面积精度来看,Planet影像的光谱特征+植被指数+均值特征结合面向对象RF分类,总体精度和Kappa系数分别为93.77%和0.87,面积的平均误差最低为4.76%;(3)采用Planet和Sentinel-2影像结合面向对象分类提取玉米倒伏精度高于基于像元分类。研究不仅分析了面向对象方法的优势,还评估了使用不用影像数据结合面向对象方法的适用性,可以为遥感提取作物倒伏相关研究提供一定的借鉴。     

Object-oriented Detection of Landslides based on the Spectral，Spatial and Morphometric Properties of Landslides

Earthquakes in mountain area often induce hundreds of thousands of landslides resulting in destructive casualties and economic damage.It is urgent needed to rapidly detect the extent areas of the landslides.With the advent of very high resolution satellite remote sensing,the application of the object\|oriented classification method in this area have significant advantage comparing to those of visual interpretation and pixel\|based methods.However,the study of object\|oriented landslide detection is relatively few,and the study usually has a small study area.The method of object\|oriented rapid identification of landslides based on the spectral,spatial and morphometric properties of landslides and a 2.5m SPOT5 multi\|spectral image is proposed in this paper and is applied in a relatively large study area.The normalized difference vegetation index (NDVI) threshold was set to remove vegetation objects and obtain landslide candidates.Then,the spectral characteristics,texture,terrain features and context of the image were used to build indicators to gradually separate the landslide from false positives.The small scale chessboard segmentation was conducted to further eliminate vegetation objects and get the landslide objects.The object\|oriented detection results show that the adopted method can recognize about 95% of the landslides in the study area.When considering the landslide excessive detection and omissions,the landslide detection quality percentage of the proposed method is 74.04%.Hence,the method proposed in the article can help to rapid assess landslide disasters caused by earthquakes or heavy rainfalls,providing a reference for post\|disaster emergency relief and reconstruction work.     

应用高分辨率遥感影像提取作物种植面积

利用中低分辨率遥感影像提取作物分类种植面积的精度,往往难以满足农业遥感估产的需要。随着新型传感器的不断出现,应用高分辨率遥感影像高精度地提取作物分类面积日益成为发展趋势。由于高分辨率遥感影像提供的地物纹理、色调与形状等信息更加丰富,当前基于对象的地物识别分类方法仍不成熟,处理操作中人为干预过多,而且较为复杂,因此尝试以地面调查信息为辅助参量,采用常规基于像元的最大似然法监督分类方法,依据多尺度遥感影像信息提取的原理,分阶段地逐步提取作物种植面积,以此为农业遥感估产服务。     

Study on Extraction Method of Abandoned Farmland based on the Seasonal Variation Characteristics of Remotely Sensed Images

In southwestern China, the cultivation conditions are poor, the plots are relatively fragmented, and the types of plots are complex. Therefore, the use of low and medium resolution remote sensing data is not able to satisfy the needs of abandoned farmland extraction. This paper explored the ability of single or multi-phased high resolution remotely sensed images in detecting abandoned farmland in southwest China, using Xiuwen County, Guizhou Province, China as a case study area. Remote sensing based monitoring methods for abandoned farmland were developed, providing a reference for the statistical survey of abandoned farmland in southwest China.The extraction method of abandoned farmland was proposed based on the field survey data, considering different types of abandoned farmland. Sensitive feature sets of different types of abandoned farmland were identified from a series of features including the spectral characteristics, vegetation indices and multi-temporal difference vegetation indices. The CART decision tree classification method was applied on the selected sensitive features to extract abandoned farmland. The results showed that:(1) There was a significant difference in the recognition ability of single-phase image in extracting different types of abandoned farmland, so it was difficult to use only single-phase image to extract abandoned farmland with high accuracy; (2) The vegetation index change characteristics of different time phases had strong recognition ability for abandoned farmland, and the ratio vegetation index was better than the difference vegetation index and normalized vegetation index; (3) The spatial distribution map of abandoned farmland and the statistical analysis of abandoned farmland area were carried out in Xiuwen County, Guizhou Province. The area of abandoned farmland in Xiuwen County was about 6,460 hectares, accounting for 13% of the cultivated land area.（4）Based on multi-temporal high-resolution remote sensing data, the method of detecting abandoned farmland using seasonal variation characteristics can meet the requirements of high-precision extraction of abandoned farmland in southwest China, and the results provided technical reference for remote sensing survey and mapping of abandoned farmland in large-scale.     

多时态BIT遥感图像建筑物的变化检测

针对来自相同地理空间的不同时刻遥感图像之间的季节性和光度变化(色差)等因素所引起的干扰, 提出了多时态-BIT遥感图像变化检测方法. 该方法引入了过去多个不同时刻的遥感图像, 融合当前遥感图像与过去时态遥感图像两两变化检测的结果, 该方法有助于排除季节性和光度变化引起的误报, 提高了变化检测的准确性; 并且利用过去多个不同时刻的遥感图像, 进一步消除非目标建筑变化的影响, 其变化点像素差值引入作为损失函数正则化项, 从而进一步提高变化检测的鲁棒性和可靠性. 本文以三时态(3个不同时刻的遥感图像)为例, 使用了遥感图像建筑物变化数据集进行了实验. 实验结果表明, 多时态-BIT方法相对于仅考虑两个时态的变化检测方法, 在遥感图像建筑物变化检测任务中表现出更好的效果.     

面向对象多特征融合的水域岸线目标变化检测

为认真落实河长制"清四乱"等专项行动,量化水域岸线监管测评工作,以岸线码头为目标,研究一种基于面向对象思想多特征融合的水域岸线目标变化检测方法。针对多时相高分辨率遥感影像,利用面向对象多尺度分割原理将具有空间连续性的同类区域划分为目标对象,提取目标对象的光谱、纹理及几何结构组成特征矩阵,并利用高斯径向基核函数支持向量机(RBF-SVM)进行分类;计算变化矢量差值,并与人工判读数据对比分析得到目标变化检测结果。实验结果表明,该研究应用于水域岸线上目标的变化检测中效果明显,RBFSVM分类误差影响最终目标变化检测的正确率,可为实现河湖水域岸线长效管护提供技术支撑。     

基于季相变化特征的撂荒地遥感提取方法研究

在我国西南地区耕种条件差,地块比较破碎,地块类型比较复杂,中低分辨率遥感数据难以满足撂荒地提取的需要。选取贵州修文县为试验区,基于高分辨率卫星遥感数据（哨兵2号）,探索单期或多期影像在中国西南地区的撂荒地检测能力,构建撂荒地遥感监测方法,为今后我国西南地区撂荒地统计调查提供参考。结合野外调查数据,在划分不同撂荒地类型基础上,综合遥感影像的光谱特征、植被指数特征以及多时相植被指数变化特征分析,优选不同类别撂荒地遥感提取敏感特征集,利用CART决策树分类方法,提取不同类型的撂荒地。结果表明：①单个时相对不同类型的撂荒地识别能力差异显著,基于单时相影像,难以开展撂荒地高精度遥感监测提取;②不同时相的植被指数变化特征对撂荒地的识别能力较强,其中比值植被指数优于差值植被指数和归一化植被指数;③以贵州修文县为例,开展了撂荒地空间分布制图及撂荒面积统计分析,修文县撂荒地面积约为6 460 hm²,占修文县耕地面积的13%;④基于多时相高分辨遥感数据,通过季相变化特征构建的撂荒地检测方法,能够满足我国西南地区撂荒地高精度遥感监测提取,为大范围撂荒地遥感调查和制图提供技术参考。     

The area extraction of winter wheat in mixed planting area based on Sentinel-2 a remote sensing satellite images

ABSTRACT

The traditional area extraction method mainly depends on manual field survey methods, it is workload, slow and high cost. While remote sensing technology has the advantages of accuracy, rapidity, macroscopic and dynamic, which has become an effective means to extract crop growing area. In this paper, we took Kaifeng City in Henan Province as the study area. Firstly, we explored the advantages of Sentinel-2A RENDVI in crop identification. Then used the supervised classification SVM, object-oriented classification method and assisted with field measured data to extract the winter wheat planting area, the characteristics of the two methods were compared and analysed. Finally, we combined the above two classification methods and proposed a new classification method V2OAE to remove unnecessary influencing factors. The experiment results showed that RENDVI has better recognition ability than the NDVI (Normalized Difference Vegetation Index) in distinguishing vegetation with similar spectrum, the classification effect of object-oriented classification is better than supervised classification SVM, and our classification method removes unnecessary influence factors in the results of object-oriented classification, which is further improve the monitoring accuracy.

Firstly, we have preprocessed the Sentinel-2A image data, its steps are: (1) In the first step, we made radiation calibration for remote sensing images to eliminate the image distortion caused by external factors, data acquisition and transmission systems and so on; (2) In the second step, we made atmospheric correction to eliminate changes in the spectral feature of remote sensing images caused by atmospheric absorption or scattering; (3) In the third step, we made band resampling to unify the resolution of remote sensing images and facilitate the mathematical combination operation of vegetation index; (4) In the fourth step, we made mosaic and cutting to get preprocessed remote sensing images of Kaifeng City. Secondly, we analysed the spectral features of each object and established the interpretation mark with the field measured data. then we explored the ability to identify the ground objects based on NDVI(Normalized Difference Vegetation Index) and RENDVI. Third, we used the rule-based object-oriented classification method and SVM classification to extract the planting area of the study area, the input definition of SVM is spectral feature images of ground objects and the output definition of SVM is the recognition result of ground objects in the process of data training. Then the advantages and disadvantages of the two methods in classification results were analysed. Finally, In order to extract winter wheat information more accurately, we combined the above two classification methods and proposed a new classification method V2OAE (Vector Object Oriented Area Extraction) to remove unnecessary influencing factors, then the winter wheat planting area in Kaifeng City was statistically obtained.