基于形态学梯度的高光谱图像分类研究

基于传统的SVM理论,首先获取像素的形态学梯度信息,考虑周围邻域的影响,对原始的梯度进行中值滤波,然后基于滤波后的梯度进行SVM分类。分类结果表明,基于空间相关性的、梯度的SVM分类精度高于基于像素灰度值的SVM分类精度。     

一种基于卷积神经网络的SAR变化检测方法

为了降低合成孔径雷达(SAR)影像中相干斑对变化检测的影响、减少标注样本的人工成本,该文发展了一种联合分层模糊C均值聚类(FCM)与卷积神经网络的非监督SAR变化检测方法。首先,利用邻域均值比算子计算前后时相的差异图,并利用分层FCM将差异图非监督地初始分割为变化类、非变化类及待确定类别像素;然后,为解决非监督选取样本时出现的样本不均衡问题,提出一种频率不变降采样的数据抽样方法,选取高置信度的变化与非变化样本用于网络训练;最后,利用训练完成的神经网络对待确定类别像素进行分类,得到最终变化结果。采用真实SAR影像数据进行实验。结果表明,该文方法方便有效,具有较高的检测精度。     

一种针对精化样本的非监督SAR变化检测方法

针对非监督框架下的深度学习SAR变化检测方法在样本选取时出现样本非平衡及冗余问题,改进了一种针对精化样本的非监督SAR影像变化检测方法:利用分层FCM针对不同差异图进行分割确定初始类别;提出利用邻域清理规则进行潜在错误样本剔除并利用异质类空间距离完成对样本的筛选;构造卷积神经网络完成对待确定类别像素的分类,得到最终变化结果.采用三组实验数据进行实验,结果表明该文方法可行有效,且能在较好应对样本问题的同时具有较高的变化检测精度.     

面向类型特征的自适应阈值遥感影像变化检测

为了实现两个不同年份单时相遥感影像之间的土地覆盖变化检测，提出了一种基于土地覆盖类型特征自适应确定阈值的遥感影像变化检测方法。以2015年土地覆盖数据为基础，综合2013年和2015年Landsat 8-OLI影像数据，首先，采用时相不变点群法TIC（Temporally Invariant Cluster）保证了两期影像辐射水平的一致性。其次，对两期影像进行多尺度分割，并在各级尺度下构建分割对象的变化向量。然后，采用最大类间方差的方法分别进行单一变化阈值变化检测以及基于土地覆盖类型的多阈值变化检测分析，并利用目视解译样点进行精度验证与评价。结果表明：（1）单一阈值变化检测结果的总体精度为79.6%，Kappa系数为0.601，多阈值变化检测结果的总体精度为87.2%，Kappa系数为0.741，多阈值变化检测具有更高的精度。（2）进一步逐土地覆盖类型精度评价可知，多阈值变化检测能在一定程度上减弱物候期的影响，具有更高的稳定性。该研究以土地覆盖数据为底图，逐类别的选取变化检测阈值，提高了变化区域检测的精度，在大范围高效更新土地覆盖数据的应用中具有一定的参考价值。     

顾及空间邻域关系的多时相SAR影像变化检测

提出了一种顾及空间邻域关系的多时相SAR影像非监督变化检测方法，采用马尔可夫随机场（Markov random field，MRF）模型描述SAR比值差异图像的空间上下文信息，提出了基于该模型的EMMPM非监督变化检测算法。实例研究表明，与未顾及空间上下文信息的EM双阈值算法相比。该方法能够有效地提高变化区域提取的可靠性和准确性。     

纹理特征向量与最大化熵法相结合的SAR影像非监督变化检测

合成孔径雷达(SAR)影像具有明显的斑点噪声,在变化检测中,一般需要考虑空间邻域信息。本文结合SAR影像丰富的纹理信息,提出一种考虑空间邻域信息的高分辨率SAR影像非监督变化检测方法,用基于灰度共生矩阵(GLCM)的32维纹理特征向量构造差异影像。通过最大化熵法自动选取阈值,对精度指标随窗口大小的变化进行回归分析,得到适合于变化检测的窗口为11×11。试验表明,本文方法优于马尔科夫随机场法,可以减小斑点噪声的影响,有效提高高分辨率SAR影像变化检测的精度。     

结合邻域信息和结构特征的遥感影像变化检测

为提高像元级变化检测方法的精度,提出一种结合邻域信息和结构特征的遥感影像变化检测方法.该方法涵盖邻域相关影像(neighborhood correlation image,NCI)、匹配误差和结构特征3种属性特征.首先,通过邻域相关分析技术获得表示上下文信息的邻域相关影像,利用邻域间像素的互相关性进行模板匹配获得匹配误差.然后,基于方向梯度信息提取能抵抗影像间光谱差异的结构特征.随后将邻域相关影像、匹配误差、结构特征作为决策树的分类属性,获取初始变化检测结果.最后,利用马尔可夫随机场(Markov random field,MRF)对其进行优化,获得最终的二值变化图.本文通过采用两组不同传感器的双时相遥感影像进行试验.结果表明,相较于采用变化向量分析法(change vector analysis,CVA)、单一邻域信息法及邻域信息和纹理特征相结合的方法,本文方法有效提高了变化检测的精度.     

基于随机森林特征优选的冬小麦分类方法

本文基于多时相Landsat 8 OLI数据,进行综合光谱、植被指数的特征提取与特征选择的方法研究。通过分析光谱与植被指数特征时序变化,提取最佳时相光谱,构建小麦提取特征;采用基于重要性与Pearson相关性的随机森林特征选择算法优选特征。结果表明:利用优选特征分类时,总体精度为89.78%,小麦分类精度为98.33%;与优选前特征的分类结果相比,精度分别提高了2.96%、2.55%;基于重要性与Pearson相关性的随机森林特征选择提高了分类精度和分类器工作效率。     

一种面向地理对象的遥感影像变化检测方法

根据高空间分辨率影像上变化区域呈聚集状分布的特点,提出了一种面向地理对象的遥感影像变化检测算法。在利用Mean-Shift分割算法的基础上,获得不同时相地理对象的灰度特征信息,结合变化矢量分析,采用最大数学期望算法自动提取变化区域。以QuickBird、SPOT、TM三组不同空间分辨率的影像进行算法验证并比较了该方法与单像素变化检测算法的差异。结果表明,三组影像中面向对象的变化检测算法的检测精度分别为91.1%,87.3%和84.3%,单像素的变化检测算法检测精度分别为86.41%,82.48%和81.02%。试验结果显示面向对象的算法检测精度高于基于单像素的变化检测算法,且对高空间分辨率的影像检测效果要优于对中低空间分辨率的影像的检测效果。该算法减少了变化阈值确定中的人工干预,克服了以像素为单位的变化检测算法中由于缺少空间邻域信息而产生孤立、离散、不连通变化结果的问题,能够满足在不同土地覆盖类型下的变化检测要求,在国土资源监测中具有一定的使用价值。     

粒子群算法及多阈值指数熵的遥感影像变化检测方法研究

在遥感影像检测中,一般采用多变化阈值来提高检测精度,但会导致运算量的增加。为解决该问题,提出利用粒子群算法及多阈值指数熵的遥感影像变化检测新方法。首先采用影像差值法构造差异影像;然后提出利用粒子群和多阈值指数熵的遥感影像分割方法,并将其用于对差异影像进行分割获取变化区域;最后对选取的实验数据进行变化检测,并与基于模糊C均值、双阈值指数熵、三阈值指数熵的非监督变化检测方法进行比较。实验结果显示,提出的变化检测方法其精度为94.77%,本案方法是一种有效地、可靠的遥感影像变化检测方法。     

综合聚类和上下文特征的高光谱影像分类

常规高光谱影像逐像素分类往往没有考虑空间相关性,分类结果未体现地物的空间关联和分布特征。为了在分类中充分利用空间特征,利用聚类信息并结合隐马尔可夫随机场模型讨论了高光谱遥感影像光谱-空间分类方法。首先,在不同特征提取方法（最小噪声分离、独立成分分析和主成分分析）下,使用不同聚类方法（k-均值、迭代自组织分析算法和模糊c-均值算法）借助隐马尔可夫随机场获取优化的分割图;然后,采用4连通区域标记法对分割区域标记生成图像对象,并根据支持向量机的逐像素分类结果采用多数投票法对图像对象进行分类;最后,借助凹槽窗口邻域滤波技术改进分类结果,削弱“椒盐”现象。该方法综合了监督分类和非监督分类的优势,通过聚类引入地物空间相关性信息,通过隐马尔可夫随机场引入上下文特征,较好地弥补了单纯基于光谱信息分类的不足。     

Development of fuzzy rule-based parameters for urban object-oriented classification using very high resolution imagery

Urban areas consist of spectrally and spatially heterogeneous features. Advanced information extraction techniques are needed to handle high resolution imageries in providing detailed information for urban planning applications. This study was conducted to identify a technique that accurately maps impervious and pervious surfaces from WorldView-2 (WV-2) imagery. Supervised per-pixel classification algorithms including Maximum Likelihood and Support Vector Machine (SVM) were utilized to evaluate the capability of spectral-based classifiers to classify urban features. Object-oriented classification was performed using supervised SVM and fuzzy rule-based approach to add spatial and texture attributes to spectral information. Supervised object-oriented SVM achieved 82.80% overall accuracy which was the better accuracy compared to supervised per-pixel classifiers. Classification based on the proposed fuzzy rule-based system revealed satisfactory output compared to other classification techniques with an overall accuracy of 87.10% for pervious surfaces and an overall accuracy of 85.19% for impervious surfaces.     

日光温室的纹理特征分层提取

针对高分辨率影像上日光温室的信息提取问题,该文提出了利用支持向量机、最近邻算法结合纹理特征在不同层上分别提取连片日光温室和独栋日光温室的方法。实验表明:纹理特征能提高分类精度,在大尺度的层上,分类精度提升幅度较大,但在小尺度的层上,分类精度提升幅度会比较小;并不是参与运算特征数越多,分类精度越高,多数情况下光谱+纹理组合的分类精度最高;提取连片日光温室的最优方案是支持向量机和光谱+形状+纹理(7像素×7像素),总精度为92.86%,Kappa系数为0.90,而提取独栋日光温室最优方案为SVM和光谱+纹理(11像素×11像素),总精度为88.39%,Kappa系数为0.86。     

面向线性光谱混合分解的邻域像元集螺线型构建方法

高时间分辨率遥感影像在地表景观破碎区域易形成混合像元,难以发挥其高时间维度优势。现有方式多是基于线性光谱混合模型,借助邻域像元所构成的像元集合组成线性方程组,求出组分光谱值的最小二乘解,提高其空间分辨率。然而,现有方法依赖窗口形式来构建邻域像元集合,在某些区域易造成方程组无解的欠定问题。本文在分析其问题原因的基础上,引入阿基米德螺线代替传统的矩形窗口,对邻域各像元依次遍历,构建空间邻近、组分相近的邻域像元集合来解决该问题。在GlobeLand 30数据上的试验表明,螺线型构建方法对5种混合尺度上多种类型地物均具有稳定的精度,与传统窗口构建方法相比,可从构建邻域像元集合方面将总体理论精度提高2%,分解结果精度提高近1个数量级。     

GF-2影像面向对象典型城区地物提取方法

国产高分遥感影像信息丰富,提供了精准的地物空间细节,深入研究高分数据处理及其提取城区地类目标信息的方法具有重要意义。本文以国产高分二号（GF-2）遥感影像为数据源,利用规则集的面向对象分类方法,通过ESP尺度分析工具选取得出最优分割尺度,建立各类地物的特征体系及分类规则,最终提取出研究区典型城区地物信息,并将之与传统基于像元的SVM监督分类结果作比较。结果表明：规则集的面向对象分类总体精度为92.23%,Kappa系数为0.9,比SVM监督分类有大幅度提高。对高分二号等高分辨率影像,面向对象的分类方法精度更高,图示效果更好,是城区地物提取的有效方法。     

Vegetation mapping with multitemporal NDVI in North Eastern China Transect (NECT)

Temporal changes in the normalized difference vegetation index (NDVI) have been widely used in vegetation mapping due to the usefulness of NDVI data in distinguishing characteristic seasonal differences in the phenology of greenness of vegetation cover. Research has also shown that NDVI provides potential to derive meaningful metrics that describe ecosystem functions. In this paper, we have applied both unsupervised “k-means” classification and supervised minimum distance classification as derived from temporal changes in NDVI measured in 1997 along the North Eastern China Transect (NECT), and we have also utilized the same two classification methods together with NDVI-derived metrics, namely maximum NDVI, mean NDVI, NDVI amplitude, NDVI threshold, total length of growing season, fraction of growing season during greenup, rate of greenup, rate of senescence, integrated NDVI during the growing season, and integrated NDVI during greenup/integrated NDVI during senescence to map vegetation. The main objectives of this study are: (1) to test the relative performance of NDVI temporal profile metrics and NDVI-derived metrics for vegetation cover discrimination in NECT; (2) to test the relative performance of unsupervised (k-means) and supervised (minimum distance) methods for vegetation mapping; (3) to test the accuracy of the IGBP-DIS released land cover map for NECT; (4) to provide an up-to-date vegetation map for NECT. The results suggest that the classifications based on NDVI temporal profile metrics have higher accuracies than those based on any other metrics, such as NDVI-derived metrics, or all (NDVI temporal profile metrics + NDVI-derived metrics), or 15 metrics (NDVI temporal profile + Rate of greenup, Rate of senescence, and Integrated NDVI in greenup/integrated NDVI in senescence) for both methods. And among them, unsupervised k-means classification had the highest overall accuracy of 52% and Kappa coefficient of 0.2057. Both unsupervised (k-means) and supervised (minimum distance) methods achieved similar accuracies for the same metrics. The accuracy of IGBP-DIS released land cover map had an overall accuracy of 37% and a Kappa coefficient is 0.1441, and can improve to 46% by decomposing the crop/natural vegetation mosaic to cropland and other natural vegetation types. The results support using unsupervised k-means classification based on NDVI temporal profile metrics to provide an up-to-date vegetation cover classification. However, new effort is necessary in the future in order to improve the overall performance on this issue.     

Comparing supervised and unsupervised multiresolution segmentation approaches for extracting buildings from very high resolution imagery

Although multiresolution segmentation (MRS) is a powerful technique for dealing with very high resolution imagery, some of the image objects that it generates do not match the geometries of the target objects, which reduces the classification accuracy. MRS can, however, be guided to produce results that approach the desired object geometry using either supervised or unsupervised approaches. Although some studies have suggested that a supervised approach is preferable, there has been no comparative evaluation of these two approaches. Therefore, in this study, we have compared supervised and unsupervised approaches to MRS. One supervised and two unsupervised segmentation methods were tested on three areas using QuickBird and WorldView-2 satellite imagery. The results were assessed using both segmentation evaluation methods and an accuracy assessment of the resulting building classifications. Thus, differences in the geometries of the image objects and in the potential to achieve satisfactory thematic accuracies were evaluated. The two approaches yielded remarkably similar classification results, with overall accuracies ranging from 82% to 86%. The performance of one of the unsupervised methods was unexpectedly similar to that of the supervised method; they identified almost identical scale parameters as being optimal for segmenting buildings, resulting in very similar geometries for the resulting image objects. The second unsupervised method produced very different image objects from the supervised method, but their classification accuracies were still very similar. The latter result was unexpected because, contrary to previously published findings, it suggests a high degree of independence between the segmentation results and classification accuracy. The results of this study have two important implications. The first is that object-based image analysis can be automated without sacrificing classification accuracy, and the second is that the previously accepted idea that classification is dependent on segmentation is challenged by our unexpected results, casting doubt on the value of pursuing ‘optimal segmentation’. Our results rather suggest that as long as under-segmentation remains at acceptable levels, imperfections in segmentation can be ruled out, so that a high level of classification accuracy can still be achieved.     

An Adaptive Supervised Nonlinear Feature Extraction for Hyperspectral Imagery Classification

In this paper, an improved version of locally linear Embedding is proposed. In the proposed method, spectral correlation angle is invited to describe the distance between data points, which is expected to fit the hyperspectral image (HSI). The neighborhood graph of the data points is constructed based on supervised method. Different from traditional supervised feature extraction methods, the weight factors, which are used to control the transform, are adaptively achieved. In this way, the input arguments of original algorithm are not increased. To justify the effectiveness of the proposed method, experiments are conducted on two HSIs. Results show that the proposed method can improve the separability of HSI especially in low dimensions.     

构建时空融合模型进行水稻遥感识别

传统变化检测手段进行水稻遥感识别受"云污染"和影像间配准误差导致的变化检测误差累积及"椒盐"现象的影响,水稻遥感识别精度低。本文提出时空融合模型(Temporal-Spatial-Fusion Model,TSFM)进行水稻遥感识别,旨在综合像元在时间、空间维度上的信息定义像元的水稻时空归属度,根据时空归属度划分阈值提取水稻。实验结果表明:在不同窗口尺度下,TSFM在整体和"云污染"区域对水稻提取均达到了较高精度。当窗口尺度为3×3时,水稻提取的用户精度、制图精度和总体精度分别达到93.4%、83.5%和87.9%。在不同窗口尺度下水稻提取的用户精度、制图精度、总体精度均高于分类后比较PCC(Post-Classification Comparison)和多数投票法(Majority Voting,MV);在"云污染"区域,水稻识别总体精度均在92.0%以上,水稻制图精度比PCC、MV分别至少提高了14.0%、7.6%。有效地解决了传统变化检测作物遥感识别存在的误差累积问题,在一定程度上避免了"云污染"和"椒盐"现象对识别结果的影响。另外,初步探讨了TSFM水稻提取精度与景观特征关系,发现在景观规整区域适宜采用较小的窗口,在破碎区域适宜采用较大的窗口。该方法的成功实施,为大范围开展秋粮作物遥感识别,消除"云"影响进行了前期实验探讨。