遥感图像森林类型小波纹理的SVM法分类

利用遥感图像对森林类型进行分类是大面积地调查、监测、分析森林资源的快速与经济的方法,但由于不同森林的光谱特征非常相近而较难准确分类。因此,在GPS数据和高分辨率遥感图像的支持下,对水源林Landsat TM遥感图像用窗口法获得阔叶林、针叶林和竹林样本图像,然后计算其小波分解后小波系数的l1范数纹理测度构成分类特征向量,利用支持向量基SVM进行分类。结果表明,利用SVM对图像中阔叶林、针叶林和竹林分类平均精度在80%以上,可较准确地识别森林类型,图像总体分类精度达到90.2%,Kappa系数0.77,均比利用小波纹理特征的神经网络法和最大似然法有所提高,森林分类错误产生的主要原因是混交林造成两类森林间存在交集。该方法可以较有效地提高遥感图像森林类型的分类精度。     

遥感图像森林林型SVM分类的多特征选择

为了研究遥感图像森林林型SVM分类多特征的选择对提高分类精度的影响,选取小波变换不同尺度纹理、四种植被指数、最优波段光谱特征等不同组合构成林型分类多特征向量进行分类。结果表明,纹理与植被指数、最优波段组合多特征的森林林型分类精度最高,阔叶林、针叶林和竹林的分类精度分别为84.4%、86.5%、91.0%,比纹理单类特征分类分别提高4.1%、4.0%、1.1%,比植被指数单类特征分类分别提高9.2%、11.8%、11.9%。多特征的分类精度一般要高于单类特征,纹理能够较明显提高林型可分性,植被指数也有一定的效果,但最优波段光谱特征的效果较弱。     

ICA结合纹理特征的SVM盐渍化信息提取研究

以渭干河-库车河三角洲绿洲为例,利用ETM+数据,探讨了该绿洲盐渍化土地覆盖信息的提取方法。提出了基于ICA与纹理特征的SVM复合的分类方法（简称ICA-T-SVM法）,通过此方法对该绿洲进行分类研究,并将分类结果与基于ICA的SVM法（简称ICA-SVM）、单源数据（光谱）SVM法、最大似然法（MLC）、神经网络法（Neural Network）分类结果进行定性和定量比较分析。研究结果表明：该方法能够有效地解决单数据源分类效果破碎、分类精度不高等问题,并对高纬输入向量具有较高的推广能力。总精度达到93.418 3%,比基于ICA的SVM法提高了3.412 3%,比单源信息的SVM分类法提高了3.423 7%,比最大似然法提高了4.979 6%,比神经网络法提高了7.714 4%,取得了良好的效果。与传统的分类方法的比较表明,文中所提出的分类方法具有明显的优越性和良好的前景,因此该方法更适合于遥感图像分类和盐渍化信息提取,是地物遥感信息提取的有效途径。     

人工智能技术在遥感图像分类中的应用

研究遥感图像分类精度问题,遥感图像分类根据图像特征进行分类,然而其特征维数相当高且信息冗余严重,分类器不能降低特征维数,导致分类器计算量大,图像分类效率和正确率低。利用主成分分析(PCA)降维特征维数的优点,提出一种基于PCA-SVM的遥感图像分类方法。PCA-SVM算法首先采用LBP算子提取遥感图像特征,然后采用PCA对遥感图像特征进行降维处理,减少特征维数并消除特征冗余信息,获得对分类结果贡献大的特征,最后采用SVM进行遥感图像分类。仿真结果表明,PCA-SVM提高了遥感图像分类效率和正确率。     

支撑向量机在高光谱遥感图像分类中的应用

高光谱遥感图像具有维数高的特点,当样本较少时,利用传统的统计识别方法分类,分类精度低.可支撑向量机(SVM)能解决小样本、高维、非线性分类问题.采用归一化法对原始图像做预处理,再分析不同的SVM核函数对分类精度的影响;并把SVM与最小距离法,马氏距离法等的分类结果进行比较.结果表明SVM的核函数类型对分类正确率影响不大,其分类精度高于传统的统计识别方法.     

基于K型支持向量机的遥感图像分类新算法

为了提高遥感图像的分类精度和识别速度,提出了一种基于K型支持向量机（SVM）的遥感图像分类新算法,该算法将灰度共生矩阵提取的纹理特征与光谱特征相结合进行分类。对两组Landsat ETM+数据进行分类仿真实验,结果表明,在多光谱遥感图像的分类中,新算法提高了分类效率、分类精度和泛化能力,K型SVM是一种优于径向基函数SVM的分类器。     

新疆干旱区绿洲土壤盐渍化信息提取对比研究

在遥感影像分类的过程中非光谱特征起着重要的辅助作用。纹理特征作为一种重要的非光谱特征对于遥感影像分类精度的提高也有很重要的作用。以渭干河-库车河三角洲绿洲为例,利用ETM+数据,探讨了该绿洲盐渍化土地覆盖信息的提取方法。提出了基于SVM的光谱和纹理两种信息复合的分类方法,通过此方法对该绿洲进行分类研究,并将分类结果与最小距离法、最大似然法(MLC)、神经网络法(Neural net)和单源数据(光谱)SVM分类结果进行定性和定量比较分析。研究结果表明：该方法能够有效地解决单数据源分类效果破碎、分类精度不高等问题,并对高维输入向量具有较高的推广能力。总精度达到93.179 5%,比单源信息的SVM分类法提高了3.161 8%,比最大似然法提高了4.825 2%,比神经网络法提高了7.475 6%,而与最小距离法相比,总精度甚至提高了11.102 9%,取得了良好的效果。与传统的分类方法的比较表明,文中所提出的分类方法具有明显的优越性和良好的前景,因此该方法更适合于遥感图像分类和盐渍化信息提取,是地物遥感信息提取的有效途径。     

SAR图像盐渍地分类研究

以渭干河——库车河三角洲绿洲为例,利用SAR数据,采用不同的分类方法来提取该研究区盐渍化土地覆盖信息。首先用Enhanced frost滤波算法对SAR图像进行去噪处理。然后基于灰度共生矩阵理论提取去噪后的SAR图像4种纹理特征,并在不同窗口大小下筛选出有效的纹理特征。最后结合纹理特征分别采用最大似然分类法和SVM分类法对SAR图像进行分类。研究结果表明:基于纹理特征的SVM分类方法,能够有效解决单源数据信息图像分类效果破碎问题;13×13窗口的总精度达到98.2456%,Kappa系数达到0.9763,更有利于遥感图像分类和盐渍化信息监测,是地物遥感信息提取的有效途径。     

基于Landsat 8 OLI的特征变量优化提取及森林生物量反演

基于江苏省常熟市虞山地区Landsat 8OLI影像和55块调查样地数据,利用多元逐步回归法建立森林生物量模型,并讨论了预测结果及其精确性。选择包括各波段灰度值、不同波段灰度值之间的线性和非线性组合(包括18种植被指数)、纹理信息以及主成分分析、最小噪声分离变换等在内的53个特征变量。通过分析53个特征变量与森林地上、地下生物量的Pearson相关性,进行特征变量的优化提取。结果表明:所有样地无区分分析时,地上和地下生物量的模型精度均达到0.4以上,基于3种森林类型(针叶林、阔叶林和混交林)进行地上和地下生物量建模时精度有明显提高,达到0.67以上,地上生物量和地下生物量的估测结果均为混交林优于阔叶林,阔叶林优于针叶林。     

利用SVM与灰度共生矩阵从QuickBird影像中提取枇杷信息

以福建省莆田市东圳水库库区为例,采用QuickBird卫星影像,利用主成分分析方法对灰度共生矩阵方法提取的地物纹理特征进行筛选,选择最佳的影像纹理特征,组成新的波段组合,并应用支持向量机方法(Support Vector Machine,SVM)进行枇杷树的提取分类,最后与只依靠光谱信息来分类的SVM法分类结果进行比较,其分类总精度由原来的71.33%提高到了86.67%,Kappa系数也由原来的0.6410提高到了0.8293,分类精度明显提高,表明光谱信息加入纹理特征信息能辅助并提升高分辨率遥感枇杷树信息提取的精度。     

结合LiDAR单木分割和高光谱特征提取的城市森林树种分类

基于多源遥感数据的城市森林树种分类对城市森林资源调查、森林健康状况评价及科学化管理具有重要意义。以江苏省常熟市虞山国家森林公园内的典型城市森林树种为研究对象,利用同期获取的机载激光雷达(LiDAR)和高光谱数据,针对5个典型城市森林树种进行了树种分类的研究。首先,基于点云距离判断单木分割方法进行单木位置和冠幅提取,并借助实测数据和目视解译结果进行精度验证;然后,在冠幅内提取4组高光谱特征变量,并借助随机森林模型对特征变量进行重要性分析;最后,筛选出重要性高的特征变量进行2个级别的树种分类并借助混淆矩阵进行验证评价。结果表明:基于点云距离判断分割方法的单木位置提取精度较高(探测率为85.7%,准确率为96%,总体精度为90.9%);利用全部特征变量(n=36)对5个树种进行分类,分类的总体精度达到了84%,Kappa系数为0.80;利用优选特征变量(n=9)进行分类,总体精度达83%,Kappa系数为0.79;利用全部特征变量(n=36)对两种森林类型进行分类,分类的总体精度达91.3%,Kappa系数为0.82,其中阔叶树种分类精度为95.6%,针叶树种分类精度为85%;利用优选特征变量(n=9)进行分类,分类的总体精度达90.7%,Kappa系数为0.80,其中阔叶树种分类精度为93.33%,针叶树种分类精度为86.67%。     

Annual seasonality in Sentinel-1 signal for forest mapping and forest type classification

ABSTRACT

The Sentinel-1 satellites provide the formerly unprecedented combination of high spatial and temporal resolution of dual polarization synthetic aperture radar data. The availability of dense time series enables the derivation and analysis of temporally filtered annual backscatter signals. The study concentrates on the use of Sentinel-1 seasonal backscatter signatures for forest area estimation and forest type classification. A classification method based on time series similarity measures is introduced and tested in three test areas covered by various forest types including broadleaf temperate, boreal and montane forests. The results are compared with two European-wide Copernicus high resolution layers, namely forest type and tree cover density (TCD). The correspondence of forest/non-forest maps and TCD is high in all test areas, with overall accuracies for forest/non-forest classification between 86% and 91% and Pearson correlation coefficients for TCD between 0.68 and 0.74. The forest type classification (non-forest, coniferous and broadleaf forest classes) provides best results in temperate forests with an overall accuracy of 85%; in boreal forest, the accuracy decreases to only 65%. Generally, the method provides reliable results for forest area estimation, including regions where methods based on static parameters are often problematic (mountainous areas), and it enables forest type classification in temperate forests.     

Study on Mangrove Species Classification based on ZY-3 Image

A hybrid mangrove forest extraction and species classification model for large coastal region was proposed using a ZY-3 (ZiYuan-3) image in the eastern part of Beibu Gulf (located at the junction of Guangdong and Guangxi).Firstly,the coastline was extracted according to the spectral features of ZY-3 image.Secondly,the buffer zone along with the coastline was generated as the suitable area of mangrove distribution.Mangrove forests and non-mangrove areas were then further classified using threshold method based on object-based image classification in these areas.Finally,Mangrove forests were classified at specie level using three pixel-based supervised classification methods,k-Nearest Neighbor,Bayes,and Random Forest.The classification results and accuracies were also compared and discussed.The results indicated that object-based threshold method can extract the coastline accurately and map the mangrove forests effectively.The pixel-based random forest classifier can obtain satisfactory results (the overall accuracy of 82.24%) of mangrove species classification than the other classifiers.In summary,the hybrid mode proposed in this paper is suitable for mangrove forests mapping and species classification in a large region.It is also validated the feasibility application of ZY-3 image in coastal mangrove research.     

Incorporating texture into classification of forest species composition from airborne multispectral images

Although research with digital airborne remote sensing data has been undertaken in different ecoregions to classify forested areas, the potential role of such imagery in deriving information to assist forest management has not yet been fully defined. The objective of this study was to determine the extent that the addition of texture could improve spectral classification of high spatial resolution images (pixel size 1m). These images represented pure and mixed wood forest stands from ecoregions in Alberta and New Brunswick, Canada. This study employed a judicious, selective application of texture to stands within a hierarchical classification framework. In Alberta, the addition of texture made a modest improvement in classification accuracy from 60% to 65%. In New Brunswick, the application of texture to selected land cover types resulted in an overall 12% improvement in classification accuracy. The addition of image texture increased classification accuracy for high spatial detail imagery relative to low spatial detail imagery. Incorporating texture into classification also improved classification accuracies for hardwood stands more so than for softwood stands, but greater attention to stand structure and composition will be needed in future work. Classification accuracies on the order of 60-65% were achieved with simple texture derivatives, maximum likelihood decision rules and conventional classification methods.     

基于高分辨遥感数据的胡杨与柽柳树冠提取

胡杨、柽柳是干旱荒漠区生境的指示种,其树冠提取是荒漠生境遥感定量监测的基础。以塔里木河下游胡杨、柽柳为研究对象,基于QuickBird数据,使用光谱单数据源SVM、光谱结合纹理SVM、面向对象分类和最大似然分类法提取树冠。结果表明:1光谱结合纹理SVM比光谱单源SVM分类精度高9.65%,冠幅估测精度高7.18%,表明高分辨影像上纹理是提高分类精度的重要因素;2面向对象分类法精度最高,分类总体精度86.47%,较光谱单源SVM提高15.67%,较光谱结合纹理SVM提高6.02%,较最大似然法提高22.58%,其冠幅估测精度达87.45%。它兼顾面向对象影像分割与支持向量机方法优点,有效利用分割对象光谱、纹理和空间等信息,较好地解决了其他方法"同物异谱、异物同谱"造成提取树冠破碎的问题,使树冠提取具有较好的稳定性和较高精度。     

Mixed KPCA结合纹理特征的SVM盐碱土信息提取

核函数是核主成分分析（Kernel Principal Component Analysis,KPCA）的核心,目前使用的核函数都是单一核函数。尝试通过将光谱角径向基核函数（Spectral Angle Radial Basis Function,SA-RBF）与RBF组合形成混合核函数。在研究中,利用基于该混合核函数的KPCA进行特征提取,将其光谱特征波段和纹理特征相结合用于盐碱土的SVM分类,将分类结果与其他SVM分类进行比较,结果表明：该方法优于其他SVM方法,能有效提取玛纳斯河流域绿洲区的盐碱土专题信息,分类精度是89.000%,kappa系数是0.876。     

The multispectral separability of Costa Rican rainforest types with support vector machines and Random Forest decision trees

Estimating the extent of tropical rainforest types is needed for biodiversity assessment and carbon accounting. In this study, we used statistical comparisons to determine the ability of Landsat Thematic Mapper (TM) bands and spectral vegetation indices to discriminate composition and structural types. A total of 144 old-growth forest plots established in northern Costa Rica were categorized via cluster analysis and ordination. Locations for palm swamps, forest regrowth and tree plantations were also acquired, making 11 forest types for separability analysis. Forest types classified using support vector machines (SVM), a theoretically superior method for solving complex classification problems, were compared with the random forest decision tree classifier (RF). Separability comparisons demonstrate that spectral data are sensitive to differences among forest types when tree species and structural similarity is low. SVM class accuracy was 66.6% for all forest types, minimally higher than the RF classifier (65.3%). TM bands and the Normalized Difference Vegetation Index (NDVI) combined with digital elevation data notably increased accuracies for SVM (84.3%) and RF (86.7%) classifiers. Rainforest types discriminated here are typically limited to one or two categories for remote sensing classifications. Our results indicate that TM bands and ancillary data combined via machine learning algorithms can yield accurate and ecologically meaningful rainforest classifications important to national and international forest monitoring protocols.     

Satellite-based detection of bamboo expansion over the past 30 years in Mount Tianmushan,China

The present work aims to detect bamboo expansion and its impact on carbon storage in a thick forest in the most recent 30 years. The research area is the national nature reserve of Tianmushan, Zhejiang Province, China, and the present paper monitored bamboo expansion from 1984 to 2015. Multi-spectral band and vegetation indices from Landsat images in summer and winter are used combined to improve the accuracy of detection using a support vector machine (SVM) classifier. Expansion of bamboo over this period is evident. Total expansion is 161%, the fastest annual rate being 11.6%. However, over recent decades the growth of bamboo has been inhibited by human activity and the total area has decreased by 21%. Evergreen broadleaf forest is the most vulnerable to invasion by bamboo at a ratio of about 65%, and this expanding trend has been brought under effective control. Carbon storage was estimated using sample plot surveys and modelling based on key ecological forests. According to our estimation using carbon storage models, the total carbon storage of Tianmushan has declined by circa 4.7% due to bamboo expansion in the past three decades.