A NEW UNSUPERVISED CLASSIFICATION ALGORITHM FOR POLARIMETRIC SAR IMAGES BASED ON FUZZY SET THEORY

In this letter, a new method is proposed for unsupervised classification of terrain types and man-made objects using POLarimetric Synthetic Aperture Radar (POLSAR) data. This technique is a combi-nation of the usage of polarimetric information of SAR images and the unsupervised classification method based on fuzzy set theory. Image quantization and image enhancement are used to preprocess the POLSAR data. Then the polarimetric information and Fuzzy C-Means (FCM) clustering algorithm are used to classify the preprocessed images. The advantages of this algorithm are the automated classification, its high classifica-tion accuracy, fast convergence and high stability. The effectiveness of this algorithm is demonstrated by ex-periments using SIR-C/X-SAR (Spaceborne Imaging Radar-C/X-band Synthetic Aperture Radar) data.

Region-based classification by combining MS segmentation and MRF for POLSAR images

Speckle effects on classification results can be sup- pressed to some extent by introducing the contextual information. An unsupervised classification algorithm is proposed for polarimetric synthetic aperture radar （POLSAR） images based on the mean shift （MS） segmentation and Markov random field （MRF）. First, polarimetdc features are exacted by target decomposition for MS segmentation. An initial classification is executed by using the target decomposition and the agglomerative hierarchical clus- tering algorithm. Thereafter, a classification step based on MRF is performed by using the mean coherence matrices obtained for each segment. Under the MRF framework, the smoothness term is defined according to the distance between neighboring areas. By using POLSAR images acquired by the German Aerospace Centre and National Aeronautics and Space Administration/Jet Propulsion Laboratory, the experimental results confirm that the proposed method has higher accuracy and better regional connectivity than other classification methods.

Intertidal area classification with generalized extreme value distribution and Markov random field in quad-polarimetric synthetic aperture radar imagery

Classification of intertidal area in synthetic aperture radar(SAR) images is an important yet challenging issue when considering the complicatedly and dramatically changing features of tidal fluctuation. The difficulty of intertidal area classification is compounded because a high proportion of this area is frequently flooded by water, making statistical modeling methods with spatial contextual information often ineffective. Because polarimetric entropy and anisotropy play significant roles in characterizing intertidal areas, in this paper we propose a novel unsupervised contextual classification algorithm. The key point of the method is to combine the generalized extreme value(GEV) statistical model of the polarization features and the Markov random field(MRF) for contextual smoothing. A goodness-of-fit test is added to determine the significance of the components of the statistical model. The final classification results are obtained by effectively combining the results of polarimetric entropy and anisotropy. Experimental results of the polarimetric data obtained by the Chinese Gaofen-3 SAR satellite demonstrate the feasibility and superiority of the proposed classification algorithm.

Double Polarization SAR Image Classification based on Object-Oriented Technology

This paper proposed to use double polarization synthetic aperture radar (SAR) image to classify surface feature, based on DEM. It takes fully use of the polarization information and external information. This pa-per utilizes ENVISAT ASAR APP double-polarization data of Poyang lake area in Jiangxi Province. Com-pared with traditional pixel-based classification, this paper fully uses object features (color, shape, hierarchy) and accessorial DEM information. The classification accuracy improves from the original 73.7% to 91.84%. The result shows that object-oriented classification technology is suitable for double polarization SAR’s high precision classification.

Integration of SAR Polarimetric Features and Multi-spectral Data for Object-Based Land Cover Classification

An object-based approach is proposed for land cover classification using optimal polarimetric parameters.The ability to identify targets is effectively enhanced by the integration of SAR and optical images.The innovation of the presented method can be summarized in the following two main points:①estimating polarimetric parameters(H-A-Alpha decomposition)through the optical image as a driver;②a multi-resolution segmentation based on the optical image only is deployed to refine classification results.The proposed method is verified by using Sentinel-1/2 datasets over the Bakersfield area,California.The results are compared against those from pixel-based SVM classification using the ground truth from the National Land Cover Database(NLCD).A detailed accuracy assessment complied with seven classes shows that the proposed method outperforms the conventional approach by around 10%,with an overall accuracy of 92.6%over regions with rich texture.

New algorithm of target classification in polarimetric SAR

The different approaches used for target decomposition （TD） theory in radar polarimetry are reviewed and three main types of theorems are introduced： those based on Mueller matrix, those using an eigenvector analysis of the coherency matrix, and those employing coherent decomposition of the scattering matrix. Support vector machine （SVM）, as a novel approach in pattern recognition, has demonstrated success in many fields. A new algorithm of target classification, by combining target decomposition and the support vector machine, is proposed. To conduct the experiment, the polarimetric synthetic aperture radar （SAR） data are used. Experimental results show that it is feasible and efficient to target classification by applying target decomposition to extract scattering mechanisms, and the effects of kernel function and its parameters on the classification efficiency are significant.

SAR images classification method based on Dempster-Shafer theory and kernel estimate

To study the scene classification in the Synthetic Aperture Radar （SAR） image, a novel method based on kernel estimate, with the Maxkov context and Dempster-Shafer evidence theory is proposed. Initially, a nonpaxametric Probability Density Function （PDF） estimate method is introduced, to describe the scene of SAR images. And then under the Maxkov context, both the determinate PDF and the kernel estimate method axe adopted respectively, to form a primary classification. Next, the primary classification results are fused using the evidence theory in an unsupervised way to get the scene classification. Finally, a regularization step is used, in which an iterated maximum selecting approach is introduced to control the fragments and modify the errors of the classification. Use of the kernel estimate and evidence theory can describe the complicated scenes with little prior knowledge and eliminate the ambiguities of the primary classification results. Experimental results on real SAR images illustrate a rather impressive performance.

A statistical distribution texton feature for synthetic aperture radar image classification

We propose a novel statistical distribution texton(s-texton) feature for synthetic aperture radar(SAR) image classification. Motivated by the traditional texton feature, the framework of texture analysis, and the importance of statistical distribution in SAR images, the s-texton feature is developed based on the idea that parameter estimation of the statistical distribution can replace the filtering operation in the traditional texture analysis of SAR images. In the process of extracting the s-texton feature, several strategies are adopted, including pre-processing, spatial gridding, parameter estimation, texton clustering, and histogram statistics. Experimental results on Terra SAR data demonstrate the effectiveness of the proposed s-texton feature.

Contrast Enhancement Method of POLSAR Images Based on Target Decomposition Theory

Improving the target-clutter ratio （TCR） of moving targets in synthetic aperture radar （SAR）, imagery is very important for target detection and identification. In this paper, using the Cloude＇s decomposition theory, an average eovarianee matrix can be decomposed into a summation of matrices representing three different scattering processes： the single bounce scattering, double bounce scattering, and diffuse scattering. A new idea of using the combination of the three components to enhance the contrast of an image is proposed. In order to compare with the polarimetric contrast enhancement method based on HH, HV, and W data, ship areas of two combinatorial intensity images are detected by image binarization. Experimental results show that the method proposed in this paper provides better contrast.

Modified version of three-component model-based decomposition for polarimetric SAR data

A new hybrid Freeman/eigenvalue decomposition based on the orientation angle compensation and the various extended volume models for polarimetric synthetic aperture radar(PolSAR) data are presented. There are three steps in the novel version of the three-component model-based decomposition.Firstly, two special unitary transform matrices are applied on the coherency matrix for deorientation to decrease the correlation between the co-polarized term and the cross-polarized term.Secondly, two new conditions are proposed to distinguish the manmade structures and the nature media after the orientation angle compensation. Finally, in order to adapt to the scattering properties of different media, five different volume scattering models are used to decompose the coherency matrix. These new conditions pre-resolves man-made structures, which is beneficial to the subsequent selection of a more suitable volume scattering model.Fully PolSAR data on San Francisco are used in the experiments to prove the efficiency of the proposed hybrid Freeman/eigenvalue decomposition.

Novel polarimetric SAR speckle filtering algorithm based on mean shift

For better interpretation of synthetic aperture radar（SAR） images,the speckle filtering is an important issue.In the area of speckle filtering,the proper averaging of samples with similar scattering characteristics is of great importance.However,existing filtering algorithms are either lack of a similarity judgment of scattering characteristics or using only intensity information for similarity judgment.A novel polarimetric SAR（PolSAR） speckle filtering algorithm based on the mean shift theory is proposed.As polarimetric covariance matrices or coherency matrices form Riemannian manifold,the pixels with similar scattering characteristics gather closely and those with different scattering characteristics separate in this hyperspace.By using the range-spatial joint mean shift theory in Riemannian manifold,the pixels chosen for averaging are ensured to be close not only in scattering characteristics but also in the spatial domain.German Aerospace Center（DLR） L-Band Experiment SAR（E-SAR） data and East China Research Institute of Electronic Engineering（ECRIEE） PolSAR data are used to demonstrate the efficiency of the proposed algorithm.The filtering results of two commonly used speckle filtering algorithms,refined Lee filtering algorithm and intensity driven adaptive neighborhood（IDAN） filtering algorithm,are also presented for the comparison purpose.Experiment results show that the proposed speckle filtering algorithm achieves a good performance in terms of speckle filtering,edge protection as well as polarimetric characteristics preservation.

Dry/wet snow mapping based on the synergistic use of dual polarimetric SAR and multispectral data

We propose a multi-sensor multi-spectral and bi-temporal dual-polarimetric Synthetic Aperture Radar(SAR) data integration scheme for dry/wet snow mapping using Sentinel-2 and Sentinel-1 data which are freely available to the research community. The integration is carried out by incorporating the information retrieved from ratio images of the conventional method for wet snow mapping and the multispectral data in two different frameworks. Firstly, a simple differencing scheme is employed for dry/wet snow mapping, where the snow cover area is derived using the Normalized Differenced Snow Index(NDSI). In the second framework, the ratio images are stacked with the multispectral bands and this stack is used for supervised and unsupervised classification using support vector machines for dry/wet snow mapping. We also investigate the potential of a state of the art backscatter model for the identification of dry/wet snow using Sentinel-1 data. The results are validated using a reference map derived from RADARSAT-2 full polarimetric SAR data. A good agreement was observed between the results and the reference data with an overall accuracy greater than 0.78 for the different blending techniques examined. For all the proposed frameworks, the wet snow was better identified. The coefficient of determination between the snow wetness derived from the backscatter model and the reference based on RADARSAT-2 data was observed to be 0.58 with a significantly higher root mean square error of 1.03 % by volume.

Polarimetric entropy of the ocean surface with a two-scale scattering model

The relationships among an ocean wave spectrum,a fully polarimetric coherence matrix,and radar parameters are deduced with an electromagnetic wave theory.Furthermore,the relationship between the polarimetric entropy and ocean wave spectrum is established based on the definition of entropy and a twoscale scattering model of the ocean surface.It is the first time that the polarimetric entropy of the ocean surface is presented in theory.Meanwhile,the relationships among the fully polarimetric entropy and the parameters related to radar and ocean are discussed.The study is the basis of further monitoring targets on the ocean surface and deriving oceanic information with the entropy from the ocean surface.The contrast enhancement between human-made targets and the ocean surface with the entropy is presented with quad-pol airborne synthetic aperture radar（AIRSAR） data.

Spectral Clustering with Eigenvalue Similarity Metric Method for POL-SAR Image Segmentation of Land Cover

A simple and fast approach based on eigenvalue similarity metric for Polarimetric SAR image segmentation of Land Cover is proposed in this paper. The approach uses eigenvalues of the coherency matrix as to construct similarity metric of clustering algorithm to segment SAR image. The Mahalanobis distance is used to metric pairwise similarity between pixels to avoid the manual scale parameter tuning in previous spectral clustering method. Furthermore, the spatial coherence constraints and spectral clustering ensemble are employed to stabilize and improve the segmentation performance. All experiments are carried out on three sets of Polarimetric SAR data. The experimental results show that the proposed method is superior to other comparison methods.

结合极化白化滤波和SimSD-CapsuleNet的PolSAR图像配准

极化合成孔径雷达(PolSAR)图像配准在地物分类、变化检测、图像融合中都具有广泛应用。现有的PolSAR图像配准方法,无论是基于深度学习还是传统方法,大多采用PolSAR幅度影像信息进行处理。这种处理方式导致大量极化信息丢失,同时在PolSAR图像固有相干斑噪声影响下,配准精度和可靠性表现不佳。为此,本文首先发展了一种有效的基于极化白化滤波(PWF)精细化处理的关键点检测器,利用PWF对PolSAR图像进行相干斑噪声抑制,通过阈值约束、形态学腐蚀及非极大值抑制来选取显著且分布均匀的匹配关键点。进一步地,本文设计了一种孪生简单稠密胶囊网络(SimSD-CapsuleNet)来快速提取PolSAR图像的浅层纹理特征和深层语义特征,同时为了充分利用极化信息,本文将极化协方差矩阵作为输入数据。本文计算了胶囊形式特征描述符之间的距离,并将其输入硬L2损失函数用于模型的训练。本文方法在不同传感器获取的不同分辨率PolSAR图像上进行验证。结果表明,该方法能够在更短的时间内获取更加均匀且数量更多的匹配关键点,结合PWF和深度神经网络可以实现快速准确的PolSAR图像配准。

基于复数域Transformer-Unet混合模型的PolSAR地物分类

传统的基于深度学习的极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,PolSAR)地物分类方法,通过堆叠卷积层提取图像局部特征,难以建立长距离依赖关系。基于自注意力机制的深度学习模型Transformer (变换)在图像分类任务中取得了成功,其自注意力机制能够捕获全局像素之间的关联性,同时PolSAR地物分类任务被证实:相比于实数域,其在复数域上表现出更好的分类效果。因此,本文将Transformer引入到复数域中,提出了一种基于复数域的Transformer和Unet (语义分割网络)混合模型(CT-Unet)用于PolSAR地物分类,将Transformer与CNN相结合,对复数类型的PolSAR数据进行特征提取,使用西安数据集和德国数据集进行PolSAR地物分类的实验结果表明:提出的模型能够有效提高PolSAR地物分类的准确性,Transformer有望在PolSAR地物分类任务中弥补卷积神经网络的不足。

PolSAR有源假目标干扰的鉴别与对消

本文针对分时极化测量体制,研究了极化合成孔径雷达(PolSAR)有源假目标干扰的鉴别与对消问题.首先建立了真实雷达目标和有源假目标干扰的极化信号模型,指出了真实雷达目标和有源假目标干扰的极化特性差异.在此基础上,提出了一种在慢时间多普勒域进行有源假目标干扰鉴别和对消的原理和方法,并给出了抗有源假目标干扰的PolSAR成像的工程实现流程.仿真实验验证了该方法的有效性.

基于极化G0分布和MRF的多视PolSAR图像迭代分类方法

提出了一种多视极化合成孔径雷达(PolSAR)图像的迭代分类方法。首先利用极化G0分布描述多视极化协方差矩阵的统计特性,并进行初始的最大似然分类,然后利用马尔可夫随机场(MRF)估计像素类标号的先验概率,最后根据MAP(最大后验概率)准则对PolSAR图像进行分类。整个分类流程迭代进行。分类结果表明该方法精度高,收敛速度快。利用NASA/JPL获取的4视AIRSAR实测数据验证了本文方法的有效性。