地球观测1号高光谱与全色图像融合的最佳方法

受制于成像原理及制造技术等因素,航天高光谱遥感图像的空间分辨率相对较低,为此提出将高光谱图像与高空间分辨率图像进行融合处理,设计最佳的增强高光谱遥感图像空间分辨率的融合算法。针对地球观测1号(EO-1)Hyperion高光谱图像和高级陆地成像仪（ALI）全色波段图像的特点,从9种具体遥感图像融合算法中选用4种融合算法开展山区与城市的数据融合实验,即Gram-Schmidt光谱锐化融合法、平滑调节滤波(SFIM)变换融合法、加权平均法（WAM）融合法和小波变换（WT）融合法,并分别从定性、定量和分类精度三方面对这些方法的融合效果进行综合评价与对比分析,从而确定适合EO-1高光谱与全色图像融合的最佳方法。实验结果显示：从图像融合效果看,在所采用的4种融合方法中,Gram-Schmidt光谱锐化融合法的效果最好;从图像分类效果看,基于融合图像的分类效果要优于基于源图像的分类效果。理论分析与实验结果均表明：Gram-Schmidt光谱锐化融合法是一种较为理想的高光谱与高空间分辨率遥感图像的融合算法,为提高高光谱遥感图像的清晰度、可靠性及图像的地物识别和分类的准确性提供有力的支持。     

基于DS聚类的高光谱图像集成分类算法

针对高光谱遥感图像维数高、样本少导致分类精度低的问题,提出一种基于DS聚类的高光谱图像集成分类算法(DSCEA)。首先,根据高光谱数据特点,从整体波段中随机选择一定数量的波段,构成不同的训练样本;其次,分析图像的空谱信息,构造无向加权图,利用优势集(DS)聚类方法得到最大特征差异的波段子集;最后,根据不同样本,利用支持向量机训练具有差异的单个分类器,采用多数表决法集成最终分类器,实现对高光谱遥感图像的分类。在Indian Pines数据集上DSCEA算法的分类精度最高可达到84.61%,在Pavia University数据集上最高可达到91.89%,实验结果表明DSCEA算法可以有效的解决高光谱分类问题。     

Mapping invasive plants using hyperspectral imagery and Breiman Cutler classifications (randomForest)

Invasive nonindigenous plants are threatening the biological integrity of North American rangelands, as well as the economies that are supported by those ecosystems. Spatial information is critical to fulfilling invasive plant management strategies. Traditional invasive plant mapping has utilized ground-based hand or GPS mapping. The shortfalls of ground-based methods include the limited spatial extent covered and the associated time and cost. Mapping vegetation with remote sensing covers large spatial areas and maps can be updated at an interval determined by management needs. The objective of the study was to map leafy spurge (Euphorbia esula L.) and spotted knapweed (Centaurea maculosa Lam.) using 128-band hyperspectral (5-m and 3-m resolution) imagery and assess the accuracy of the resulting maps. Beiman Cutler classifications (BCC) were used to classify the imagery using the randomForest package in the R statistical program. BCC builds multiple classification trees by repeatedly taking random subsets of the observational data and using random subsets of the spectral bands to determine each split in the classification trees. The resulting classification trees vote on the correct classification. Overall accuracy was 84% for the spotted knapweed classification, with class accuracies ranging from 60% to 93%; overall accuracy was 86% for the leafy spurge classification, with class accuracies ranging from 66% to 93%. Our results indicate that (1) BCC can achieve substantial improvements in accuracy over single classification trees with these data and (2) it might be unnecessary to have separate accuracy assessment data when using BCC, as the algorithm provides a reliable internal estimate of accuracy.     

The role of digital bathymetry in mapping shallow marine vegetation from hyperspectral image data

Hyperspectral remote sensing is a proven technology for measurement of coastal ocean colour, including sea‐bed mapping in optically shallow waters. Using hyperspectral imagery of shallow (<15 m deep) sea bed acquired with the Compact Airborne Spectrographic Imager (CASI‐550), we examined how changes in the spatial resolution of bathymetric grids, created from sonar data (echosounding) and input to conventional image classifiers, affected the accuracy of distributional maps of invasive (Codium fragile ssp. tomentosoides) and native (kelp) seaweeds off the coast of Nova Scotia, Canada. The addition of a low‐resolution bathymetric grid, interpolated from soundings by the Canadian Hydrographic Service, improved the overall classification accuracies by up to ～10%. However, increasing the bathymetric resolution did not increase the accuracy of classification maps produced with the supervised (Maximum Likelihood) classifier as shown by a slightly lower accuracy (2%) when using an intermediate‐resolution bathymetric grid interpolated from soundings with a recreational fish finder. Supervised classifications using the first three eigenvectors from a principal‐components analysis were consistently more accurate (by at least 27%) than unsupervised (K‐means classifier) schemes with similar data compression. With an overall accuracy of 76%, the most reliable scheme was a supervised classification with low‐resolution bathymetry. However, the supervised approach was particularly sensitive, and variations in accuracy of 2% resulted in overestimations of up to 53% in the extent of C. fragile and kelp. The use of a passive optical bathymetric algorithm to derive a high‐resolution bathymetric grid from the CASI data showed promise, although fundamental differences between this grid and those created with the sonar data limited the conclusions. The bathymetry (at any spatial resolution) appeared to improve the accuracy of the classifications both by reducing the confusion among the spectral classes and by removing noise in the image data. Variations in the accuracy of depth estimates and inescapable positional inaccuracies in the imagery and ground data largely accounted for the observed differences in the classification accuracies. This study provides the first detailed demonstration of the advantages and limitations of integrating digital bathymetry with hyperspectral data for the mapping of benthic assemblages in optically shallow waters.     

On combining multiscale deep learning features for the classification of hyperspectral remote sensing imagery

In recent years, satellite imagery has greatly improved in both spatial and spectral resolution. One of the major unsolved problems in highly developed remote sensing imagery is the manual selection and combination of appropriate features according to spectral and spatial properties. Deep learning framework can learn global and robust features from the training data set automatically, and it has achieved state-of-the-art classification accuracies over different image classification tasks. In this study, a technique is proposed which attempts to classify hyperspectral imagery by incorporating deep learning features. Firstly, deep learning features are extracted by multiscale convolutional auto-encoder. Then, based on the learned deep learning features, a logistic regression classifier is trained for classification. Finally, parameters of deep learning framework are analysed and the potential development is introduced. Experiments are conducted on the well-known Pavia data set which is acquired by the reflective optics system imaging spectrometer sensor. It is found that the deep learning-based method provides a more accurate classification result than the traditional ones.     

The effect of spectral and spatial degradation of hyperspectral imagery for the Sclerophyll tree species classification

Hyperspectral imaging can be a useful remote-sensing technology for classifying tree species. Prior to the image classification stage, effective mapping endeavours must first identify the optimal spectral and spatial resolutions for discriminating the species of interest. Such a procedure may contribute to improving the classification accuracy, as well as the image acquisition planning. In this work, we address the effect of degrading the original bandwidth and pixel size of a hyperspectral and hyperspatial image for the classification of Sclerophyll forest tree species. A HySpex-VNIR 1600 airborne-based hyperspectral image with submetric spatial resolution was acquired in December 2009 for a native forest located in the foothills of the Andes of central Chile. The main tree species of this forest were then sampled in the field between January and February 2010. The original image spectral and spatial resolutions (160 bands with a width of 3.7 nm and pixel sizes of 0.3 m) were systematically degraded by resampling using a Gaussian model and a nearest neighbour method, respectively (until reaching 39 bands with a width of 14.8 nm and pixel sizes of 2.4 m). As a result, 12 images with different spectral and spatial resolution combinations were created. Subsequently, these images were noise-reduced using the minimum noise fraction procedure and 12 additional images were created. Statistical class separabilities from the spectral divergence measure and an assessment of classification accuracy of two supervised hyperspectral classifiers (spectral angle mapper (SAM) and spectral information divergence (SID)) were applied for each of the 24 images. The best overall and per-class classification accuracies (>80%) were observed when the SAM classifier was applied on the noise-reduced reflectance image at its original spectral and spatial resolutions. This result indicates that pixels somewhat smaller than the tree canopy diameters were the most appropriate to represent the spatial variability of the tree species of interest. On the other hand, it suggests that noise-reduced bands derived from the full image spectral resolution rendered the best discrimination of the spectral properties of the tree species of interest. Meanwhile, the better performance of SAM over SID may result from the ability of the former to classify tree species regardless of the illumination differences in the image. This technical approach can be particularly useful in native forest environments, where the irregular surface of the uppermost canopy is subject to a differentiated illumination.     

Performance characteristics of the indicator classifier on simulated image data

Classification of remotely sensed imagery into groups of pixels having similar spectral reflectance characteristics is conducted classically by comparing the spectral signature of unknown pixels with those of training pixels of known ground cover type. Thus classification methods use only the spectral characteristics of image data without considering the spatial aspects or the relative location of an unknown pixel with respect to pixels from the training data set. An indicator classifier was introduced in 1992 that combines spatial and spectral information in a decision model. In this Letter the performance of this classifier is tested on simulated image data with known mineral targets and varying spatial variability and noise. It is demonstrated that incorporating spatial continuity into the classification process may largely increase the accuracy of the resulting classified images.     

Object-based urban detailed land cover classification with high spatial resolution IKONOS imagery

Improvement in remote sensing techniques in spatial/spectral resolution strengthens their applicability for urban environmental study. Unfortunately, high spatial resolution imagery also increases internal variability in land cover units and can cause a ‘salt-and-pepper’ effect, resulting in decreased accuracy using pixel-based classification results. Region-based classification techniques, using an image object (IO) rather than a pixel as a classification unit, appear to hold promise as a method for overcoming this problem. Using IKONOS high spatial resolution imagery, we examined whether the IO technique could significantly improve classification accuracy compared to the pixel-based method when applied to urban land cover mapping in Tampa Bay, FL, USA. We further compared the performance of an artificial neural network (ANN) and a minimum distance classifier (MDC) in urban detailed land cover classification and evaluated whether the classification accuracy was affected by the number of extracted IO features. Our analysis methods included IKONOS image data calibration, data fusion with the pansharpening (PS) process, Hue–Intensity–Saturation (HIS) transferred indices and textural feature extraction, and feature selection using a stepwise discriminant analysis (SDA). The classification results were evaluated with visually interpreted data from high-resolution (0.3 m) digital aerial photographs. Our results indicate a statistically significant difference in classification accuracy between pixel- and object-based techniques; ANN outperforms MDC as an object-based classifier; and the use of more features (27 vs. 9 features) increases the IO classification accuracy, although the increase is statistically significant for the MDC but not for the ANN.     

Feature extraction method based on spectral dimensional edge preservation filtering for hyperspectral image classification

ABSTRACT

A large amount of spectral and spatial information contained in hyperspectral imagery has provided a great opportunity to effectively characterize and identify the surface materials of interest. Feature extraction plays a very important role for hyperspectral data classification, which can reduce noise from the original data and improve the separability of land classes. A novel feature extraction technique based on spectral dimensional edge preserving filter is proposed in this paper. A series of Gaussian filters are applied in the spatial domain of the hyperspectral image to produce the guidance image, then, the edge preserving filter which is guided by the guidance image is adopted and applied in the spectral domain of the hyperspectral data to get the feature. For the feature is produced by filtering in the spectral domain, the spectral curves of the feature are more continues, which avoids the spectral discontinuity problems result from the traditional two-dimensional spatial filter. The guidance image is obtained by filtering the original image in the spatial domain, so, the spatial and the spectral information are integrated together in the following spectral edge preserving filtering process. We carefully adjusted the parameters of the filter and applied it to different real hyperspectral remote sensing images, with the support vector machine, multinomial logistic regression, and random forest serving as the classifier, by comparing with other feature extraction methods presented in recent literature, the results indicate that the proposed methodology always has a great performance in different kinds of cases.     

Edge-preserving-based collaborative representation for spectral-spatial classification

Recently, the nearest regularized subspace (NRS) classifier and its spectral–spatial versions such as joint collaborative representation (JCR) and weighted JCR (WJCR) have gained an increasing interest in the hyperspectral image classification. JCR and WJCR average each pixel with its neighbours in a spatial neighbourhood window. The use of spatial information as averaging of pixels in a local window may degrade the classification accuracy in the neighbourhood of discontinuities and class boundaries. We propose the edge-preserving-based collaborative representation (EPCR) classifier in this article, which overcomes this problem by using the edge image estimated by the original full-band hyperspectral image. The estimated edge image is used for calculation of the weights of neighbours and also the final residuals in the collaborative representation classifier. The advantage of multiscale spatial window is also assessed in this work. Moreover, the kernelized versions of NRS and its improved versions are developed in this article. Our experimental results on several popular hyperspectral images indicate that EPCR and its kernelized version are superior to some state-of-the-art classification methods.     

基于双通道变分自编码器的高光谱图像分类

针对现有高光谱图像变分自编码器（variational autoencoder,VAE）分类算法存在空间和光谱特征利用效率低的问题,提出一种基于双通道变分自编码器的高光谱图像深度学习分类算法。通过构建一维条件变分自编码器（conditional variational autoencoder,CVAE）特征提取框架和二维循环通道条件变分自编码（channel-recurrent conditional variational autoencoders,CRCVAE）特征提取框架分别提取高光谱图像的光谱特征和空间特征,将光谱特征向量和空间特征向量叠加形成空谱联合特征向量,将联合特征送入Softmax分类器中进行分类。在Indian pines和Pavia University两种高光谱数据集上进行了分析验证,实验结果显示,与其他算法相比,提出的算法在总分类精度、平均分类精度和Kappa系数等评价指标上至少提高了3.40、2.75和3.57个百分点,结果显示提出的算法得到了最高的分类精度和更好的可视化效果。     

PCA与移动窗小波变换的高光谱决策融合分类

目的 高光谱数据具有较高的谱间分辨率和相关性,给分类处理带来了一定的困难.为了提高分类精度,提出一种结合PCA与移动窗小波变换的高光谱决策融合分类算法.方法 首先,利用相关系数矩阵对原始高光谱数据进行波段分组;然后,利用主成分分析对每组数据进行谱间降维;再根据提出的移动窗小波变换法进行空间特征提取;最后,采用线性意见池(LOP)决策融合规则对多分类器的分类结果进行融合.结果 采用两组来自不同传感器的数据进行实验,所提算法的分类精度和Kappa系数均高于已有的5种分类算法.与SVM-RBF算法相比,本文算法的分类精度高出了8％左右.结论 实验结果表明,本文算法充分挖掘了高光谱图像的谱间-空间信息,能有效提高分类正确率,在小样本情况下和噪声环境中也具有良好的分类性能.     

3D密集全卷积的高光谱地物分类算法研究

针对高光谱遥感图像训练样本较少、光谱维度较高、空间特征与频谱特征存在差异性而导致高光谱地物分类的特征提取不合理、分类精度不稳定和训练时间长等问题,提出了基于3D密集全卷积（3D-DSFCN）的高光谱图像（HSI）分类算法。算法通过密集模块中的3D卷积核分别提取光谱特征和空间特征,采用特征映射模块替换传统网络中的池化层和全连接层,最后通过softmax分类器进行分类。实验结果表明,基于3D-DSFCN的HSI分类方法提高了地物分类的准确率、增强了低频标签的分类稳定性。     

Mapping freshwater marsh species distributions using WorldView-2 high-resolution multispectral satellite imagery

Freshwater wetlands are highly diverse, spatially heterogeneous, and seasonally dynamic systems that present unique challenges to remote sensing. Maximum likelihood and support vector machine-supervised classification were compared to map wetland plant species distributions in a deltaic environment using high-resolution WorldView-2 satellite imagery. The benefits of the sensor’s new coastal blue, yellow, and red-edge bands were tested for mapping coastal vegetation and the eight-band results were compared to classifications performed using band combinations and spatial resolutions characteristic of other available high-resolution satellite sensors. Unlike previous studies, this study found that support vector machine classification did not provide significantly different results from maximum likelihood classification. The maximum likelihood classifier provided the highest overall classification accuracy, at 75%, with user’s and producer’s accuracies for individual species ranging from 0% to 100%. Overall, maximum likelihood classification of WorldView-2 imagery provided satisfactory results for species distribution mapping within this freshwater delta system and compared favourably to results of previous studies using hyperspectral imagery, but at much lower acquisition cost and greater ease of processing. The red-edge and coastal blue bands appear to contribute the most to improved vegetation mapping capability over high-resolution satellite sensors that employ only four spectral bands.     

联合快舟一号影像纹理信息的城市土地覆盖分类

仅依靠光谱信息无法满足高分辨率遥感分类的应用需求,辅之以纹理特征信息进行分类,可提高影像分类精度。利用KZ\|1卫星影像和Landsat\|8卫星影像数据,基于面向对象的影像分割法和灰度共生矩阵纹理分析法对新疆石河子市局部城区进行了地表覆盖分类实验,将不同空间分辨率的全色影像纹理信息、光谱信息构成多种影像特征组合进行分类比较研究,以选择最佳的分类特征集。结果表明:KZ-1影像能为城市区域的土地覆盖分类提供丰富的纹理信息,面向对象的影像分割可较好地利用高分辨率数据的几何结构信息实现优化的影像分割,从而提高多光谱影像的分类精度,总体分类精度为90.06%,Kappa系数为87.93%,比单纯利用光谱信息分类的总体精度提高了8.02%,Kappa系数提高了9.65%,表明KZ\|1数据可为光谱分类提供丰富的纹理信息,从而提高城市区域的土地覆盖分类精度。     

ELM-based spectral–spatial classification of hyperspectral images using extended morphological profiles and composite feature mappings

Extreme Learning Machine (ELM) is a supervised learning technique for a class of feedforward neural networks with random weights that has recently been used with success for the classification of hyperspectral images. In this work, we show that the morphological techniques can be integrated in this kind of classifiers using several composite feature mappings which are proposed for ELM. In particular, we present a spectral–spatial ELM-based classifier for hyperspectral remote-sensing images that integrates the information provided by extended morphological profiles. The proposed spectral–spatial classifier allows different weights for both spatial and spectral features, outperforming other ELM-based classifiers in terms of accuracy for land-cover applications. The accuracy classification results are also better than those obtained by equivalent spectral–spatial Support-Vector-Machine-based classifiers.     

Spatial–spectral locality constrained elastic net hypergraph for hyperspectral image clustering

Hypergraph is an effective method used to represent the contextual correlation within hyperspectral imagery for clustering. Nevertheless, how to discover the closely correlated samples to form hyperedges is the key issue for constructing an informative hypergraph. In this article, a new spatial–spectral locality constrained elastic net hypergraph learning model is proposed for hyperspectral image clustering (i.e. unsupervised classification). In order to utilize the spatial–spectral correlation among the pixels in hyperspectral images, first, we construct a locality-constrained dictionary by selecting K relevant pixels within a spatial neighbourhood, which activates the most correlated atoms and suppresses the uncorrelated ones. Second, each pixel is represented as a linear combination of the atoms in the dictionary under the elastic net regularization. Third, based on the obtained representations, the pixels and their most related pixels are linked as hyperedges, which can effectively capture high–order relationships among the pixels. Finally, a hypergraph Laplacian matrix is built for unsupervised learning. Experiments have been conducted on two widely used hyperspectral images, and the results show that the proposed method can achieve a superior clustering performance when compared to state-of-the-art methods.     

高光谱影像的BDT-SVM地物分类算法与应用

面对海量数据的特征空间高维性及训练样本的有限性,高光谱遥感影像若采用常规统计模式的分类方法难以获得较好的分类结果。因此探讨支持向量机(SVM)分类器的基本原理,针对EO-1Hyperion高光谱影像的分类特点及现有多类SVM算法所存在的训练时间长及分类精度低等问题,引入二叉决策树SVM(BDT-SVM)分类算法,并提出一种新的类间分离度定义方法及相应的客观确定二叉树结构的策略,由此生成改进的BDT-SVM算法。实验结果表明:与其他多类分类方法相比,基于改进的BDT-SVM算法的高光谱影像地物分类效果更好,总体精度达到90.96%,Kappa系数为0.89,该算法还解决了经典SVM多类分类可能存在的不可分区域问题。     

A probabilistic SVM approach for hyperspectral image classification using spectral and texture features

New hyperspectral sensors can collect a large number of spectral bands, which provide a capability to distinguish various objects and materials on the earth. However, the accurate classification of these images is still a big challenge. Previous studies demonstrate the effectiveness of combination of spectral data and spatial information for better classification of hyperspectral images. In this article, this approach is followed to propose a novel three-step spectral–spatial method for classification of hyperspectral images. In the first step, Gabor filters are applied for texture feature extraction. In the second step, spectral and texture features are separately classified by a probabilistic Support Vector Machine (SVM) pixel-wise classifier to estimate per-pixel probability. Therefore, two probabilities are obtained for each pixel of the image. In the third step, the total probability is calculated by a linear combination of the previous probabilities on which a control parameter determines the efficacy of each one. As a result, one pixel is assigned to one class which has the highest total probability. This method is performed in multivariate analysis framework (MAF) on which one pixel is represented by a d-dimensional vector, d is the number of spectral or texture features, and in functional data analysis (FDA) on which one pixel is considered as a continuous function. The proposed method is evaluated with different training samples on two hyperspectral data. The combination parameter is experimentally obtained for each hyperspectral data set as well as for each training samples. This parameter adjusts the efficacy of the spectral versus texture information in various areas such as forest, agricultural or urban area to get the best classification accuracy. Experimental results show high performance of the proposed method for hyperspectral image classification. In addition, these results confirm that the proposed method achieves better results in FDA than in MAF. Comparison with some state-of-the-art spectral–spatial classification methods demonstrates that the proposed method can significantly improve classification accuracies.     

The role of spectral resolution and classifier complexity in the analysis of hyperspectral images of forest areas

Remote sensing hyperspectral sensors are important and powerful instruments for addressing classification problems in complex forest scenarios, as they allow one a detailed characterization of the spectral behavior of the considered information classes. However, the processing of hyperspectral data is particularly complex both from a theoretical viewpoint [e.g. problems related to the Hughes phenomenon (Hughes, 1968) and from a computational perspective. Despite many previous investigations that have been presented in the literature on feature reduction and feature extraction in hyperspectral data, only a few studies have analyzed the role of spectral resolution on the classification accuracy in different application domains. In this paper, we present an empirical study aimed at understanding the relationship among spectral resolution, classifier complexity, and classification accuracy obtained with hyperspectral sensors for the classification of forest areas. We considered two different test sets characterized by images acquired by an AISA Eagle sensor over 126 bands with a spectral resolution of 4.6 nm, and we subsequently degraded its spectral resolution to 9.2, 13.8, 18.4, 23, 27.6, 32.2 and 36.8 nm. A series of classification experiments were carried out with bands at each of the degraded spectral resolutions, and bands selected with a feature selection algorithm at the highest spectral resolution (4.6 nm). The classification experiments were carried out with three different classifiers: Support Vector Machine, Gaussian Maximum Likelihood with Leave-One-Out-Covariance estimator, and Linear Discriminant Analysis. From the experimental results, important conclusions can be made about the choice of the spectral resolution of hyperspectral sensors as applied to forest areas, also in relation to the complexity of the adopted classification methodology. The outcome of these experiments are also applicable in terms of directing the user towards a more efficient use of the current instruments (e.g. programming of the spectral channels to be acquired) and classification techniques in forest applications, as well as in the design of future hyperspectral sensors.