ALOS影像数据土地覆盖分类及景观特征研究

通过马氏距离法、最大似然法、支持向量机三种途径对土地覆盖进行分类,以混淆矩阵对分类结果做精度评价,结果显示,最大似然法和支持向量机分类有较好的效果。以最大似然法为例,通过引入归一化植被指数（NDVI）、基于灰度共生矩阵的纹理特征等进行不同特征组合的分类,探讨其对分类的影响。研究表明,NDVI、对比度、均值参与分类后,对分类精度都有不同程度的提高,而三者与原始波段的结合分类精度最高。基于分类结果做景观格局定量分析。结果表明,研究区景观类型较为丰富,以耕地为主导,再加上城镇和农村聚落用地,约占到整个研究区的82%,表明景观所受的人类活动干扰和压力很大、生态风险高。因此,必须强化黑河中游绿洲荒漠区的土地利用规划和管理,适当约束耕地和聚落用地的扩张,提高土地利用效率;要加强生态保护和建设,提高景观的抗干扰能力。     

基于Sentinel-2的UNVI植被指数及性能对比研究

作物精准识别和分类是农业遥感检测的重要内容,对作物长势监测以及估产十分重要。以美国混合农业带为研究区,基于Sentinel-2时间序列影像,根据其传感器响应函数计算了针对Sentinel-2的通用归一化植被指数（Universal Normalized Vegetation Index,UNVI）,并通过两个对比实验,分析UNVI等6个指数在作物精准分类中的性能。实验一以JM（Jeffries-Matusita）距离为指标对不同作物类别之间的可分性进行分析,结果表明UNVI优于NDVI、EVI、WDRVI、NDre1和NDWI指数,在玉米和棉花、玉米和水稻、玉米和水稻的区分上,UNVI优于其他指数区分能力相当,但在其余的作物组合上如棉花和水稻,NDVI等指数则无法将其很好的区分,此时UNVI指数依然可以表现出较好的区分能力;实验二对6种时间序列指数特征分别使用随机森林和支持向量机进行作物分类,结果表明UNVI指数的总体精度和Kappa系数最高,其次是NDre1指数和WDRVI指数,EVI的总体精度和Kappa系数最低,这表明UNVI比其他6个指数更好地区分了研究区大豆、玉米、棉花和水稻等4种主要作物。综上,基于Sentinel-2时间序列的UNVI指数在进行作物分类时与其他5种遥感植被指数相比,具有较大的优势,UNVI可为农作物长势分析和作物估产研究等农业研究和应用的可选植被指数。     

SVM在多源遥感图像分类中的应用研究

在利用遥感图像进行土地利用/覆盖分类过程中,可采用以下两种途径来提高分类精度:一是通过增加有利于分类的数据源,引入地理辅助数据和归一化植被指数(NDVI)来进行多源信息融合;二是选择更好的分类方法,例如支持向量机(SVM)学习方法,由于该方法克服了最大似然法和神经网络的弱点,非常适合高维、复杂的小样本多源数据的分类。为了提高多源遥感图像分类的精度,还研究了支持向量机在遥感图像分类中模型的选择,包括多类模型和核函数的选择。分类结果表明,支持向量机比传统的分类方法具有更高的精度,尤其是基于径向基核函数和一对一多类方法的支持向量机模型更适合多源遥感图像分类,因此,基于支持向量机的多源土地利用/覆盖分类能大大提高分类精度。     

基于多元数据和不同分类算法的遥感影像信息提取及精度评价 ——以祁连山东段为例

以祁连山东段典型山地系统为研究区,通过提取研究区TM影像的主成分、各类植被指数、基于灰度共生矩阵的影像纹理特征以及研究区地形特征等数据,应用最优波段指数方法得到最优波段组合,并运用非监督分类、最大似然法、支持向量机分类法、决策树分类法对上述最优波段进行分类研究。结果表明多尺度数据挖掘有利于分类精度的提高,同时选取合适的判断标准的决策树分类方法在遥感信息提取中有比较直观意义和较高的分类精度。在上述分类方法中分类精度由高到低为决策树分类>支持向量机法>最大似然法>非监督分类法。决策树分类总体分类精度为94.50%,kappa系数为0.9122。
     

遥感影像的分类自校正方法研究

遥感图像分类是遥感图像研究的主要内容之一,分类精度高低直接关系到遥感数据的可靠性和实用性。多分类器系统可以提高单分类器分类的精度,但往往要求组成的子分类器分类误差相互独立,子分类器选择困难。支持向量机是新发展起来的一种非参数分类器,其分类原理和传统的基于统计的分类方法不同,表现出一定的独立性。为此本文尝试基于支持向量机和目前使用最广泛的最大似然法,构建一个性能高效且组合方式简单的复合分类器(称为遥感影像分类自校正方法)。同时,为了验证该分类器的性能,在北京市2006年4月27日的SPOT2图像上选择了一个研究区,分别利用最大似然法、支持向量机法和分类自校正方法进行分类对比试验。结果显示分类自校正方法的总体分类精度最高,比最大似然法和支持向量机法分别提高了4.35%和6.6%,而且各种地物类型的分类精度相对最大似然和支持向量机法都有提高。本文提出的分类自校正方法是一种性能高效且操作简单的分类方法。     

利用时序数据构建冬小麦识别矢量分析模型

鉴于作物类型识别中存在光谱特征相似的困扰,"异物同谱"问题难以有效解决,而时序归一化植被指数(Normalized Different Vegetation Index,NDVI)曲线数据能够反映作物不同时期的动态变化趋势,该文将NDVI时间序列投影到N维空间构成多维特征矢量,结合冬小麦特有的物候特征,充分利用矢量的方向和大小参量,构建冬小麦识别的矢量分析模型,模型的识别能力较强,可以充分发挥NDVI时间序列的优势。以唐山市为研究区,基于高分一号WFV(Wide Field of View)数据的高分辨率优势,构建覆盖冬小麦生长期的NDVI时间序列,采用矢量分析模型进行冬小麦识别,同时与最大似然法、马氏距离法、支持向量机法、神经网络法、最小距离法等分类方法进行对比。结果显示,后5种分类方法的Kappa系数介于0.701 8和0.790 3之间,而矢量分析模型达到了0.895 2,精度有了较大提高。该研究为冬小麦识别提取提供了新的思路,也对推动遥感农情信息调研具有一定学术和应用价值。同时,基于研究区训练样本提出了模型阈值参数自动确定的方法,为今后冬小麦自动提取奠定了基础。     

影响ETM影像土地利用/覆盖分类精度因素的研究

训练样本量、辅助数据和分类法是影响土地利用/覆盖分类精度的3个主要因素,通过找到这3个因素的最佳组合方式以提高分类精度,分别在25%、50%、75%、100%样本量下,加入NDVI、DEM和纹理均值特征作为辅助数据,比较了分类回归树、支持向量机、最大似然法3种分类法的效果,探讨了训练样本、辅助数据以及分类技术对土地利用/覆盖分类精度的影响。结果表明：支持向量机总体分类精度较高,在相同样本量和没有有效辅助数据的情况下,SVM可以获得最佳的分类结果,总体分类精度在85%以上;在进行分类时,加入NDVI和纹理均值特征使分类回归树分类精度提高了2.82%,说明该方法对有效辅助数据的加入较为敏感;在获取的训练样本集有限而可获取有效的辅助数据时,应优先考虑利用分类回归树进行土地利用/覆盖分类。     

基于SVM的高分时序冬小麦种植区提取研究

冬小麦作为我国重要的粮食作物,准确获取其空间分布情况,对农业生产管理及农情监测有重要意义。以河南省商丘市为例,利用覆盖冬小麦完整生育期的GF-1数据,计算归一化植被指数（Normalized Difference Vegetation Index, NDVI）、增强植被指数（Enhanced Vegetation Index, EVI）时间序列,结合关键生育期影像,构建不同特征量组合数据集,利用支持向量机方法进行冬小麦提取。同时采用主成分分析法对数据进行降维处理,尝试通过压缩特征集数据量来提高冬小麦提取效率。研究结果表明：EVI时序数据较NDVI能更好地描述作物的物候,提取精度皆高于NDVI,其中EVI时序数据与关键生育期影像组合提取精度最高,达到97.67%。结果表明,降维后数据并未对提取精度造成显著影响,达到压缩数据量保持提取精度的目的,为大区域作物提取提供参考价值。     

样本特征对参数/非参数分类器分类精度的影响分析

为验证理论训练数量(10~30 p)对参数分类器(如最大似然分类)、非参数分类器(如支撑向量机)的适用性以及样本特征(光谱统计、空间分布特征)对分类器分类精度的影响,选择不同规模的训练样本进行最大似然分类和支撑向量机分类,分析分类精度与样本之间的关系。实验结果表明:随着样本量的增加,最大似然、支撑向量机分类精度均随样本量增多而提高并趋于稳定,最大似然分类精度的增长速度要快于支撑向量机。MLC受样本量的影响较大,在小样本的时候(5个),分类精度不稳定,超过30个样本的时候,分类精度稳定下来;对于SVM分类器,在小样本的时候(5个),分类精度较高且稳定,因此SVM分类适合于小样本分类,不受限于理论样本量的影响。当样本量超过最小理论样本量值(30个)的时候,最大似然分类精度要优于支撑向量机,主要是由于当样本量增加后,最大似然更易于获得有效的信息量样本,而对于支撑向量机边缘信息样本的增加数量不大。研究结果为进一步优化样本进行分类打下前期的实验基础。     

基于多源信息的TM遥感图像计算机分类

本文以榆林市城区及其周边范围为实验区,以TM遥感图像的第一主成分纹理信息、归一化植被指数和MNF变换得到的四个波段为数据源,采用支持向量机方法进行分类,并与最大似然法分类和单纯利用光谱信息的基于SVM分类结果进行比较.试验结果表明,将纹理分析方法应用于图像分类中可区分光谱混淆的地类;和传统的分类方法相比,采用支持向量机技术.使用光谱与纹理特征结合的分类方法可以获得更高的分类精度.     

Research on Crop Classification Method based on Sentinel-2 Time Series Combined Vegetation Index

Time series is a widely used phenological research method. A new time series vegetation indices which takes full advantage of the red edge information of Sentinel 2 data were used for crop classification to improve the classification accuracy. The NDVI, EVI, and red edge NDVI were combined to construct a time series vegetation index image. Then, four different algorithms (support vector machine, random forest, CART decision tree and maximum likelihood) were used to classify four crops, three forest grasses, bare land, and water bodies. Among the original classification results, the random forest with the highest overall accuracy is 87.92%, and the maximum likelihood with the lowest overall accuracy is 80.07%. In the classification details, the boundaries of random forest and support vector machine are the clearest. Among the four classification results, the classification accuracy of crops is higher than other land types, just smaller than water body. The error mainly comes from the mixture of three forests. It indicates that the time series combined vegetation index is feasible and accurate for crop classification.     

Testing the red edge channel for improving land-use classifications based on high-resolution multi-spectral satellite data

The incorporation of a red edge channel in multi-spectral satellite sensors has potential for improving land-use classification, as the related electromagnetic spectrum is specifically sensitive to vegetation chlorophyll content. RapidEye is the first high-resolution multi-spectral satellite system that operationally provides a red edge channel. The objective of this study is to test the potential of the RapidEye red edge channel for improving the classification of land use, investigated at a study site west of Berlin. Based on a scene from July 2009, supervised land-use classifications were performed using different sets of spectral feature input, including and excluding red edge information. The algorithms used are support vector machine and maximum likelihood. The results indicate that the incorporation of red edge information can increase classification accuracy. The highest positive effects are observed for vegetation classes in open landscapes, e.g. for bush vegetation.     

Land Cover Extraction in the Ejina Oasis by Hyperspectral Remote Sensing

The extraction of land surface coverage is the basis of ecological environment evaluation,vegetation change analysis and regional ecological and hydrological processes.Aerial hyperspectral remote sensing has great advantage in land surface coverage extraction,such as flexible,wide coverage,high spatial resolution and high spectral resolution.Research area has landscape characteristics of vegetation,landscape fragmentation and heterogeneity in Ejina Poplar Forest National Nature Reserve.Comparison and analysis of two methods of dimension reduction based on minimum noise transform and principal component analysis,three supervised classification methods based on maximum likelihood method,support vector machine and object\|oriented classification.Land surface coverage is extracted by NDVI threshold segmentation,minimum noise transform dimensionality reduction method and maximum likelihood classification method according to the characteristics of landscape fragmentation,heterogeneity and high redundancy of hyperspectral data based on the Airborne Hyperspectral Data of Ejina oasis in the lower reaches of Heihe.The land surface coverage results overall accuracy and Kappa coefficient are 87.95% and 0.885 by random sampling based on airborne remote sensing data.The results show that the classification results of high accuracy can provide effective parameters for ecological research.     

Improvements in land use mapping for irrigated agriculture from satellite sensor data using a multi-stage maximum likelihood classification

The accuracy of conventional land use classification of irrigated agriculture from optical satellite images using maximum likelihood supervised classification was compared with a classification based on multistage maximum likelihood supervised classification. In the multistage maximum likelihood classification series of sub-classifications were carried out which included masking and/or omitting certain crops from the classifications. These series of classifications improved the identification of individual crops/land use types. The output from the optimum sub-classifications were stacked to give an overall crop types/land use map. When the multistage classification was tested against a single stage classification on a large irrigation scheme in Central Asia the final accuracy of crop/land use classification increased from 85% to 94%. Field verification confirmed the accuracy at 93.5%. These results were achieved with a single Landsat 7 Enhanced Thematic Mapper (ETM+) sensor dataset as of 2 August 1999 over an area of 38.5?km².     

Comparison of Machine Learning Methods for Land Use/Land Cover Classification in the Complicated Terrain Regions

Aiming at the characteristics of varied and complex geomorphic types,crisscross network of ravines and broken terrain in high altitude complicated terrain regions,it is very important to study and find the rapid and effective land use/land cover classification method for obtaining and timely updating of land use information.Taking the Huangshui river basin located in the transitional zone between the Loess Plateau and the Qinghai-Tibet Plateau as acasestudy area,the objective of this study is to explore a kind of effective information extraction method from comparison of four kinds machine learning methods for complicated terrain regions.based on Landsat 8 OLI satellite data,DEM and combined with various thematic features,on the basis of geographical division of the study area,artificial neural network,decision tree,support vector machine and random forest four machine learning methods for land use information extraction were used to obtain land use data,and confusion matrix was constructed to evaluate classification accuracy.The results showed that the classification accuracies of random forest and decision tree are obviously higher than those of support vector machine and artificial neural network.The random forest method has the highest classification accuracy,the overall classification accuracy is 85.65%,the Kappa coefficient is 0.84.based on the above classification,Random forest classification method was chose to further classify Landsat 8 fusion datafrom panchromatic 15 meter and multispectral 30 meter image,the overall classification accuracy is 86.49% and the Kappa coefficient is 0.85.This indicated that the random forest classification method can obtain higher classification efficiency while ensuring the classification accuracy.It is very effective for the extraction of land use information in complicated terrain regions.Data fusion can improve the classification accuracy to a certain extent.     

基于WorldView-2影像的土地利用信息提取方法对比及评价

基于2011年WorldView-2高分辨率遥感影像, 采取面向对象的分类方法和四种传统的基于像元的分类方法分别提取平潭县海坛岛中北部研究样区土地利用信息, 并以目视解译结果图为参考, 得到每种分类方法的总体分类精度, 且从数量分歧和分配分歧两方面对土地利用信息提取结果进行整体评价和单类别评价, 结果表明: (1)不同分类方法平均总体分类精度为75.00%, 其中最高的是面向对象法, 总体精度为84.25%, 分类总体精度最低的为最大似然法, 仅为62.00%. (2)面向对象分类方法具有最低的数量分歧, 为4.25%, 其次依次为神经网络法<支持向量机法<马氏距离法<最大似然法. 在分配分歧方面, 支持向量机方法其值最低, 为5.75%, 其次依次为最大似然法<神经网络法<马氏距离法<面向对象法. (3)在单类别精度评价中, 耕地的精度对影像整体分类结果影响最为显著, 其数量分歧比例大小依次为最大似然法(28.75%)>马氏距离法(21.50%)>支持向量机法(14.75%)>神经网络法(11.00%)>面向对象法(3.00%), 分配分歧比例大小依次为面向对象法(10.50%)>神经网络法(5.00%)>支持向量机法(1.50%)>最大似然法(0.50%)>马氏距离法(0.00%).     

基于多时序特征和卷积神经网络的农作物分类

近年来,以卷积神经网络为主的深度学习模型在各种遥感应用中都显示出巨大的潜力。以加州帝国郡为研究区,以Landsat 8 OLI年内时序遥感影像计算时序植被指数NDVI、EVI、RVI以及TVI,组合后输入到构建的一维卷积神经网络 模型,以实现作物的高精度精细分类。为了验证卷积模型的优越性,另搭建了基于递归神经网络及其变体的深度学习模型。结果表明：①引入其他时序特征后,能够有效地提高卷积神经网络的分类精度。NDVI+EVI+TVI+RVI组合特征总体精度和Kappa系数最高,分别是89.667 4%和0.856 0,对比NDVI时序特征总体精度和Kappa系数提高了近4%和0.6。②在与其他深度学习模型的对比中,一维卷积神经网络分类精度最高,能够从时序数据中较为准确捕捉作物时序特征信息,尽管递归神经网络被广泛应用于序列数据的研究,但分类结果要略差于卷积神经网络。实验表明在NDVI的基础上引入其他植被指数辅助,能够有效地提高分类精度。基于一维卷积神经网络的深度学习框架为长时间序列分类任务提供了一种有效且高效的方法。     

Drought assessment and monitoring through remote sensing and GIS in western tracts of Tamil Nadu,India

Drought is an insidious hazard of nature and is considered to be the most complex but least understood of all natural hazards. Large historical datasets are required to study drought and these involve complex interrelationships between climatological and meteorological data. Rainfall is an important meteorological parameter; the amount and distribution influence the type of vegetation in a region. To analyse the changes in vegetation cover due to variation in rainfall and identify the land-use areas facing drought risk, rainfall data from 1981 to 2003 were categorized into excess, normal, deficit and drought years. The Advanced Very High Resolution Radiometer (AVHRR) sensor's composite dataset was used for analysing the temporal and interannual behaviour of surface vegetation. The various land-use classes – crop land (annual, perennial crops), scrub land, barren land, forest land, degraded pasture and grassland – were identified using satellite data for excess, normal, deficit and drought years. Normalized Difference Vegetation Indices (NDVIs) were derived from satellite data for each land-use class and the highest NDVI mean values were 0.515, 0.436 and 0.385 for the tapioca crop in excess, normal and deficit years, respectively, whereas in the drought year, the groundnut crop (0.267) showed the maximum. Grassland recorded the lowest value of NDVI in all years except for the excess year. Annual crops, such as groundnut (0.398), pulses (0.313), sorghum (0.120), tapioca (0.436) and horse gram (0.259), registered comparatively higher NDVI values than the perennial crops for the normal year. The Vegetation Condition Index (VCI) was used to estimate vegetation health and monitor drought. Among land-use classes, the maximum VCI value of 92.1% was observed in onions for the excess year, whereas groundnut witnessed the maximum values of 78.2, 64.5 and 55.2% for normal, deficit and drought years, respectively. Based on the VCI classification, all land-use classes fall into the optimal or normal vegetation category in excess and normal years, whereas in drought years most of the land-use classes fall into the drought category except for sorghum, groundnut, pulses and grasses. These crops (sorghum 39.7%, groundnut 55.2%, pulses 38.5% and grassland 38.6%) registered maximum VCI values, revealing that they were sustained under drought conditions. It is suggested that the existing crop pattern be modified in drought periods by selecting the suitable crops of sorghum, groundnut and pulses and avoiding the cultivation of onion, rice and tapioca.