Assessing the performance of two unsupervised dimensionality reduction techniques on hyperspectral APEX data for high resolution urban land-cover mapping

Despite the high richness of information content provided by airborne hyperspectral data, detailed urban land-cover mapping is still a challenging task. An important topic in hyperspectral remote sensing is the issue of high dimensionality, which is commonly addressed by dimensionality reduction techniques. While many studies focus on methodological developments in data reduction, less attention is paid to the assessment of the proposed methods in detailed urban hyperspectral land-cover mapping, using state-of-the-art image classification approaches. In this study we evaluate the potential of two unsupervised data reduction techniques, the Autoassociative Neural Network (AANN) and the BandClust method – the first a transformation based approach, the second a feature-selection based approach – for mapping of urban land cover at a high level of thematic detail, using an APEX 288-band hyperspectral dataset. Both methods were tested in combination with four state-of-the-art machine learning classifiers: Random Forest (RF), AdaBoost (ADB), the multiple layer perceptron (MLP), and support vector machines (SVM). When used in combination with a strong learner (MLP, SVM) BandClust produces classification accuracies similar to or higher than obtained with the full dataset, demonstrating the method’s capability of preserving critical spectral information, required for the classifier to successfully distinguish between the 22 urban land-cover classes defined in this study. In the AANN data reduction process, on the other hand, important spectral information seems to be compromised or lost, resulting in lower accuracies for three of the four classifiers tested. Detailed analysis of accuracies at class level confirms the superiority of the SVM/Bandclust combination for accurate urban land-cover mapping using a reduced hyperspectral dataset. This study also demonstrates the potential of the new APEX sensor data for detailed mapping of land cover in spatially and spectrally complex urban areas.     

Data fusion and classifier ensemble techniques for vegetation mapping in the coastal Everglades

This study examined the applicability of data fusion and classifier ensemble techniques for vegetation mapping in the coastal Everglades. A framework was designed to combine these two techniques. In the framework, 20-m hyperspectral imagery collected from Airborne Visible/Infrared Imaging Spectrometer was first merged with 1-m Digital Orthophoto Quarter Quads using a proposed pixel/feature-level fusion strategy. The fused data set was then classified with an ensemble approach based on two contemporary machine learning algorithms: Random Forest and Support Vector Machine. The framework was applied to classify nine vegetation types in a portion of the coastal Everglades. An object-based vegetation map was produced with an overall accuracy of 90% and Kappa value of 0.86. Per-class classification accuracy varied from 61% for identifying buttonwood forest to 100% for identifying red mangrove scrub. The result shows that the framework is promising for automated vegetation mapping in the Everglades.     

Efficient paddy field mapping using Landsat-8 imagery and object-based image analysis based on advanced fractel net evolution approach

This paper proposes an efficient paddy field mapping method using object-based image analysis and a bitemporal data set acquired by Landsat-8 Operational Land Imager. In the proposed approach, image segmentation is the first step and its quality has a serious impact on the accuracy of paddy field classification. In order to improve segmentation quality, a new segmentation algorithm based on a frequently used method, fractal net evolution approach, is developed, with improvement mainly in merging criteria. In order to automate the process of scale parameter determination, an unsupervised scale selection method is utilized to determine the optimal scale parameter for the proposed image segmentation approach. After segmentation, four types of object-based features including geometric, spectral, textural, and contextual information are extracted and input into the subsequent classification procedure. By using a random forest classifier, paddy fields and nonpaddy fields are separated. The proposed image segmentation method and the final classification result are both quantitatively evaluated. Our segmentation method outperformed two popular algorithms according to three supervised evaluation criteria. The classification result with overall accuracy of 91.00% and kappa statistic of 0.82 validated the effectiveness of the proposed framework. Further analysis on feature importance indicated that spectral features made the most contribution as compared to the other three types of object-based features.     

A Multiple SVM System for Classification of Hyperspectral Remote Sensing Data

With recent technological advances in remote sensing sensors and systems, very high-dimensional hyperspectral data are available for a better discrimination among different complex land-cover classes. However, the large number of spectral bands, but limited availability of training samples creates the problem of Hughes phenomenon or ‘curse of dimensionality’ in hyperspectral data sets. Moreover, these high numbers of bands are usually highly correlated. Because of these complexities of hyperspectral data, traditional classification strategies have often limited performance in classification of hyperspectral imagery. Referring to the limitation of single classifier in these situations, Multiple Classifier Systems (MCS) may have better performance than single classifier. This paper presents a new method for classification of hyperspectral data based on a band clustering strategy through a multiple Support Vector Machine system. The proposed method uses the band grouping process based on a modified mutual information strategy to split data into few band groups. After the band grouping step, the proposed algorithm aims at benefiting from the capabilities of SVM as classification method. So, the proposed approach applies SVM on each band group that is produced in a previous step. Finally, Naive Bayes (NB) as a classifier fusion method combines decisions of SVM classifiers. Experimental results on two common hyperspectral data sets show that the proposed method improves the classification accuracy in comparison with the standard SVM on entire bands of data and feature selection methods.     

Hyperspectral Image Classification Using Relevance Vector Machines

This letter presents a hyperspectral image classification method based on relevance vector machines (RVMs). Support vector machine (SVM)-based approaches have been recently proposed for hyperspectral image classification and have raised important interest. In this letter, it is genuinely proposed to use an RVM-based approach for the classification of hyperspectral images. It is shown that approximately the same classification accuracy is obtained using RVM-based classification, with a significantly smaller relevance vector rate and, therefore, much faster testing time, compared with SVM-based classification. This feature makes the RVM-based hyperspectral classification approach more suitable for applications that require low complexity and, possibly, real-time classification.     

A Hierarchical Hybrid SVM Method for Classification of Remotely Sensed Data

The focus of this work is on developing a new hierarchical hybrid Support Vector Machine (SVM) method to address the problems of classification of multi or hyper spectral remotely sensed images and provide a working technique that increases the classification accuracy while lowering the computational cost and complexity of the process. The paper presents issues in analyzing large multi/hyper spectral image data sets for dimensionality reduction, coping with intra pixel spectral variations, and selection of a flexible classifier with robust learning process. Experiments conducted revealed that a computationally cheap algorithm that uses Hamming distance between the pixel vectors of different bands to eliminate redundant bands was quite effective in helping reduce the dimensionality. The paper also presents the concept of extended mathematical morphological profiles for segregating the input pixel vectors into pure or mixed categories which will enable further computational cost reductions. The proposed method’s overall classification accuracy is tested with IRS data sets and the Airborne Visible Infrared Imaging Spectroradiometer Indian Pines hyperspectral benchmark data set and presented.     

基于再生核Hilbert空间小波核函数支持向量机的高光谱遥感影像分类

针对支持向量机用于高光谱遥感影像分类存在的分类精度不高、参数选择困难等问题,提出一种再生核Hilbert空间的小波核.其可以逼近任意非线性函数,能够有效改进参数估计的效果,进而实现基于再生核Hilbert空间的小波核函数支持向量机(小波支持向量机).并选取北京昌平地区的国产高光谱数据operational modula...     

Parameter evaluation and optimization for multi-resolution segmentation in object-based shadow detection using very high resolution imagery

Object-based shadow detection in urban areas is an important topic in very high resolution remote sensing image processing. Multi-resolution segmentation (MRS) is an effective segmentation method, and is used for object-based shadow detection. However, several input parameters within MRS may result in unstable performance for final shadow detection; thus, the evaluation and optimization for the parameters upon the final shadow detection accuracy cannot be overlooked. In this paper, the three parameters in MRS (scale s, weight of colour w_color and weight of compactness w_compact) upon the final result of a recently proposed method, object-based shadow detection with Dempster–Shafer theory, were evaluated and optimized by sensitivity analysis and Taguchi’s method with three experimental data. Experiments show that scale s is the most sensitive parameter among the three parameters within MRS. More importantly, according to the Taguchi’s method theory, there is a very significant interaction effect between s and w_color, which cannot be overlooked. The shadow detection accuracy yielded by the optimum parameter combination in consideration of the interaction effect is higher than that only optimized by covering the main effect of single parameter in most cases.     

MRELBP特征、Franklin矩和SVM相结合的遥感图像建筑物识别方法

为了进一步提高遥感图像建筑物区域的识别精度,提出了一种基于中值稳健扩展局部二值模式(median robust extended local binary pattern,MRELBP)、Franklin矩和布谷鸟优化支持向量机(support vector machine,SVM)的分类方法。首先,通过MRELBP特征算子计算图像块的纹理特征向量,并根据Franklin矩得到形状特征向量,组合图像块的纹理特征向量和形状特征向量得到综合特征向量;然后,利用训练样本对SVM进行训练,同时由布谷鸟搜索算法对SVM的核函数参数和惩罚因子进行优化;最后,通过训练好的SVM得到建筑物区域识别结果。通过30组试验的结果表明,与基于三原色(red green blue,RGB)和SVM的分类方法、基于LBP和SVM的分类方法、基于Zernike矩和SVM的分类方法相比,本文提出的方法所识别的遥感图像建筑物区域准确度更高。     

面向对象规则和支持向量机的天宫一号高光谱影像分类

传统的高光谱分类方法通常基于单一像元的光谱或纹理特征,很少考虑地物空间结构信息与空间相关特征.本文将面向对象规则与基于像元的分类进行融合,利用对象的空间结构特征和光谱特征进行混合分类,旨在克服像元层次分类的不足.本文尝试性的提出了两种混合分类方法:(1)基于分形网络演化的多尺度分割支持向量机分类(2)基于多层分水岭分割的SVM分类,并将这两种方法应用到天宫一号高光谱数据上.结果表明:基于面向对象规则的混合分类方法有效地提高了分类精度,不仅能够改善同谱异物现象,而且解决分类结果中地物破碎的问题.     

基于波段子集的独立分量分析的特征提取的高光谱遥感影像分类

针对高光谱影像数据具有波段众多、数据量较大的特点,本文提出了一种基于波段子集的独立分量分析(ICA)特征提取的高光谱遥感影像分类的新方法。以北京昌平小汤山地区的高光谱影像为例,根据高光谱遥感影像的相邻波段的相关性进行子空间划分,在各个波段子集上采用ICA算法进行特征提取,将各个子空间提取的特征合并组成特征向量,采用支持向量机(SVM)分类器进行分类。结果表明:该方法分类精度最佳(分类精度89.04%,Kappa系数0.8605,明显优于其它特征提取方法的SVM分类,有效地提高了高光谱数据的分类精度。     

Multiple spectral similarity metrics for surface materials identification using hyperspectral data

Modern hyperspectral imaging and non-imaging spectroradiometer has the capability to acquire high-resolution spectral reflectance data required for surface materials identification and mapping. Spectral similarity metrics, due to their mathematical simplicity and insensitiveness to the number of reference labelled spectra, have been increasingly used for material mapping by labelling reflectance spectra in hyperspectral data labelling. For a particular hyperspectral data set, the accuracy of spectral labelling depends considerably upon the degree of unambiguous spectral matching achieved by the spectral similarity metric used. In this work, we propose a new methodology for quantifying spectral similarity for hyperspectral data labelling for surface materials identification. Developed adopting the multiple classifier system architecture, the proposed methodology unifies into a single framework the differential performances of eight different spectral similarity metrics for the quantification of spectral matching for surface materials. The proposed methodology has been implemented on two types of hyperspectral data viz. image (airborne hyperspectral images) and non-image (library spectra) for numerous surface materials identification. Further, the performance of the proposed methodology has been compared with the support vector machines (SVM) approach, and with all the base spectral similarity metrics. The results indicate that, for the hyperspectral images, the performance of the proposed methodology is comparable with that of the SVM. For the library spectra, the proposed methodology shows a consistently higher (increase of about 30% when compared to SVM) classification accuracy. The proposed methodology has the potential to serve as a general library search method for materials identification using hyperspectral data.     

高光谱影像光谱-空间多特征加权概率融合分类

提出了一种基于光谱-空间多特征加权概率融合的高光谱影像分类方法。首先,利用最小噪声分离(minimum noise fraction,MNF)方法对高光谱影像进行降维和特征提取,并以得到的MNF特征影像作为光谱特征,联合灰度共生矩阵(gray level co-occurrence matrix,GLCM)提取的纹理特征、基于OFC算子建立的多尺度形态学特征以及采用连续最大角凸锥(sequential maximum angle convex cone,SMACC)提取的端元组分特征,组成3组光谱-空间特征;然后利用支持向量机(support vector machine,SVM)对每一组光谱-空间特征进行分类,得到每组特征的概率输出结果;最后,建立多特征加权概率融合模型,应用该模型将不同特征的概率输出结果进行加权融合,得到最终分类结果。为了验证该方法的有效性,利用ROSIS和 AVIRIS影像进行试验,总体分类精度分别达到97.65%和96.62%。结果表明本文的方法不但较好地克服了传统基于单一特征高光谱影像分类的局限性,而且其分类效果也优于常规矢量叠加(vector stacking,VS)和概率融合的多特征分类方法,有效地改善了高光谱影像的分类结果。     

Per-pixel and object-oriented classification methods for mapping urban land cover extraction using SPOT 5 imagery

To have sustainable management and proper decision-making, timely acquisition and analysis of surface features are necessary. Traditional pixel-based analysis is the popular way to extract different categories, but it is not comparable by the achievements that can be achieved through the object-based method that uses the additional characteristics of features in the process of classification. In this paper, three types of classification were used to classify SPOT 5 satellite image in mapping land cover; Support vector machine (SVM) pixel-based, SVM object-based and Decision Tree (DT) pixel-based classification. Normalised Difference Vegetation Index and the brightness value of two infrared bands (NIR and SWIR) were used in manually developed DT classification. The classification of the SVM (pixel based) was generated using the selected groups of pixels that represent the selected features. In addition, the SVM (object based) was implemented by using radial-based function kernel. The classified features were oil palm, rubber, urban area, soil, water and other vegetation. The study found that the overall classification of the DT was the lowest at 69.87% while those of SVM (pixel based) and SVM (object based) were 76.67 and 81.25%, respectively.     

不同降维策略下的高光谱影像多特征分类

由于物体表面的空间分布通常是富有规律且局部连续的,在高光谱影像分类中应充分利用其光谱和空间信息。本文在对高光谱影像立方体进行降维处理的基础上,提出了一种联合空域和谱域信息的高光谱影像高效分类方法。首先,分别选用主成分分析(Principal Component Analysis,PCA)和正交投影波段选择(Orthogonal Projection Band Selection,OPBS)两种方法对原始高光谱数据进行预处理,获取降维后的影像数据。然后在其基础上提取扩展形态学特征(Extended Morphology Profiles,EMP)和地物表面纹理特征,组成联合光谱和纹理、形状结构特征。最后,采用支持向量机(Support Vector Machine,SVM)分类器对联合特征进行分类。针对不同真实高光谱数据集的实验结果表明,本文提出的方法运算效率高且具有令人满意的分类性能。     

利用SVM-CRF进行高光谱遥感数据分类

提出了一种改进的随机场模型SVM-CRF,它以支持向量机作为条件随机场的一阶势能项,结合了支持向量机和条件随机场的优点。采用AVIRIS高光谱遥感数据进行实验,对SVM-CRF模型进行了分析,结果表明,在分类精度上SVM-CRF优于支持向量机和传统条件随机场模型。     

Determination of chlorophyll content of small water bodies (kettle holes) using hyperspectral airborne data

This study presents an approach for chlorophyll content determination of small shallow water bodies (kettle holes) from hyperspectral airborne ROSIS and HyMap data (acquired on 15 May and 29 July 2008 respectively). Investigated field and airborne spectra for almost all kettle holes do not correspond to each other due to differences in ground sampling distance. Field spectra were collected from the height of 30–35 cm (i.e. area of 0.01–0.015 m²). Airborne pixels of ROSIS and HyMap imageries cover an area of 4 m² and 16 m² respectively and their spectra are highly influenced by algae or bottom properties of the kettle holes. Analysis of airborne spectra revealed that chlorophyll absorption near 677 nm is the same for both datasets. In order to enhance absorption properties, both airborne hyperspectral datasets were normalized by the continuum removal approach. Linear regression algorithms for ROSIS and HyMap datasets were derived using normalized average chlorophyll absorption spectra for each kettle hole. Overall accuracy of biomass mapping for ROSIS data was 71%, and for HyMap 64%. Biomass mapping results showed that, depending on the type of kettle hole, algae distribution, the ‘packaging effect’ and bottom reflection lead to miscalculations of the chlorophyll content using hyperspectral airborne data.     

面向高光谱影像分类的生成式对抗网络

为了提高高光谱影像分类精度,提出了一种基于生成式对抗网络的高光谱影像分类方法。生成式对抗网络由生成器、判别器和分类器3部分组成,其中生成器用于模拟高光谱样本的数据分布,生成特定类别的样本;判别器是一个二值分类器,用于判断输入的样本是否为真实数据;分类器用于对输入的样本进行分类。利用反向传播算法依次更新生成器、判别器和分类器的网络参数使损失函数最小,从而达到训练网络的目的。生成器和判别器能够模拟高光谱影像的样本分布来辅助训练分类器,因此能够提高高光谱影像的分类精度。分别采用Pavia大学和Salinas高光谱数据集进行分类试验,试验结果表明提出的分类方法能够在小样本条件下提高高光谱影像的分类精度。     

小样本的高光谱图像降噪与分类

在样本数目稀少情况下实现高光谱图像精细分类是个挑战性的问题。高光谱图像信噪比提高比较困难,噪声大小对分类结果有最直接的影响。利用高光谱图像相邻波段之间的相关性和相邻像素之间的相关性,提出多级降噪滤波的高光谱图像分类方法,通过改进的两阶段稀疏与低秩矩阵分解方法,去除高光谱图像中能量较高的噪声,利用主成分分析方法去除高光谱图像中能量较低的噪声,引导滤波方法去除分类结果图中的"椒盐噪声"。选取两幅真实高光谱图像进行实验,结果表明,两阶段稀疏与低秩矩阵分解法和主成分分析法两种降噪方法具有较强的互补性;引导滤波方法使得分类图更加平滑且分类精度更高。与其他光谱空间分类方法相比,本文方法分类精度更高,且在样本极少时能获得很高的分类精度。     

联合空-谱信息的高光谱影像深度三维卷积网络分类

针对高光谱影像分类高维和小样本的特点,提出一种基于深度三维卷积神经网络的高光谱影像分类方法。首先,该方法直接以高光谱数据立方体为输入,利用三维卷积操作提取高光谱数据立方体的三维空-谱特征。然后,利用残差学习构建深层网络,提取更高层次的特征表达,以提高分类精度。最后,采用Dropout正则化方法防止过拟合。利用Pavia大学、Indian Pines和Salinas 3组高光谱数据进行试验验证,结果表明,与支持向量机和现有的基于深度学习的高光谱影像分类方法相比,该方法能有效提高高光谱影像的地物分类精度。