The Jeffries–Matusita distance for the case of complex Wishart distribution as a separability criterion for fully polarimetric SAR data

In multidimensional observations, many classification algorithms (supervised or unsupervised) require the selection of optimum bands in which the classes are most distinct. The Jeffries–Matusita (JM) distance is widely used as a separability criterion for optimal band selection and evaluation of classification results. Its original form is based on the assumption of normal distribution of the data. However, in the case of the covariance/coherency matrix of synthetic aperture radar (SAR) polarimetry, the data follow the complex Wishart distribution. In this article, we calculate the JM separability criterion for the case of the complex Wishart distribution. The updated formulation is used for: (1) the estimation of the separability between classes in fully polarimetric SAR data and to evaluate two standard polarimetric SAR classification algorithms, the Wishart and the expectation maximization algorithms, and (2) the classification of fully polarimetric SAR images based on the derived JM separability for the case of complex Wishart distribution. Fully polarimetric RADARSAT-2 images over sea ice in the Canadian Arctic are used to classify different ice surfaces and open water.     

深度置信网络在极化SAR图像分类中的应用

目的 深度置信网络能够从数据中自动学习、提取特征,在特征学习方面具有突出优势。极化SAR图像分类中存在海量特征利用率低、特征选取主观性强的问题。为了解决这一问题,提出一种基于深度置信网络的极化SAR图像分类方法。方法 首先进行海量分类特征提取,获得极化类、辐射类、空间类和子孔径类四类特征构成的特征集;然后在特征集基础上选取样本并构建特征矢量,用以输入到深度置信网络模型之中;最后利用深度置信网络的方法对海量分类特征进行逐层学习抽象,获得有效的分类特征进行分类。结果 采用AIRSAR数据进行实验,分类结果精度达到91.06%。通过与经典Wishart监督分类、逻辑回归分类方法对比,表现了深度置信网络方法在特征学习方面的突出优势,验证了方法的适用性。结论 针对极化SAR图像海量特征的选取与利用,提出了一种新的分类方法,为极化SAR图像分类提供了一种新思路,为深度置信网络获得更广泛地应用进行有益的探索和尝试。     

故障特征组合选择方法

特征选择方法主要包括过滤方法和绕封方法。为了利用过滤方法计算简单和绕封方法精度高的优点,提出一种组合过滤和绕封方法的特征选择新方法。该方法首先利用基于互信息准则的过滤方法得到满足一定精度要求的子集后,再采用绕封方法找到最后的优化特征子集。由于遗传算法在组合优化问题上的成功应用,对特征子集寻优采用了遗传算法。在数值仿真和轴承故障特征选择中,采用新方法在保证诊断精度的同时,可以节省大量选择时间。组合特征选择方法有较好的寻优特征子集的能力,能够节省选择时间,具有高效、高精度的双重优点。     

一种基于信息增益及遗传算法的特征选择算法

特征选择是模式识别及数据挖掘等领域的重要问题之一。针对高维数据对象,特征选择一方面可以提高分类精度和效率,另一方面可以找出富含信息的特征子集。针对此问题,本文提出一种综合了filter模型及wrapper模型的特征选择方法,首先基于特征之间的信息增益进行特征分组及筛选,然后针对经过筛选而精简的特征子集采用遗传算法进行随机搜索,并采用感知器模型的分类错误率作为评价指标。实验结果表明,该算法可有效地找出具有较好的线性可分离性的特征子集,从而实现降维并提高分类精度。     

On the sensitivity of polarimetric SAR measurements to vegetation cover: the Coiba National Park,Panama

In this study, the sensitivity of multi-polarization synthetic aperture radar (SAR) features to vegetation cover is investigated over a test case of environmental importance: the Coiba National Park, Panama. Single-polarization intensity features and polarimetric features derived from the eigenvalue/eigenvector decomposition are analysed and their classification performance, evaluated against a reference land-cover map using a simple clustering algorithm, is contrasted with conventional optical features.

Experiments, undertaken using actual L-band full-polarimetric SAR and Landsat data, show that (a) polarimetric information plays a key role in improving the classification accuracy with some polarimetric features performing better than single-polarization and optical ones, (b) classification performance of radar features is significantly affected by incidence angles, and (c) a joint use of different radar features is expected to increase classification accuracy.     

改进的基于非局部均值的极化SAR相干斑抑制

贝叶斯形式的非局部均值模型在极化SAR图像相干斑抑制中有良好的应用,在实现抑制相干斑的同时较好地保持了边缘细节和点目标.通过分析合成孔径雷达(SAR)图像多视数据的空间统计分布,结合贝叶斯形式的非局部均值模型,得出在该模型下多视与单视SAR图像中像素间相似性度量函数一致性的结论,并对该相似性度量函数进行了修正,使之满足对称性;最后针对算法全局使用一个固定滤波参数影响滤波效果的问题,提出一种根据像素间相似程度自适应选取滤波参数的方法.实验结果验证了本文算法的有效性.     

极化SAR相干斑抑制的非局部加权最小均方误差滤波算法

目的 相干斑的存在严重影响了极化合成孔径雷达(PolSAR)的影像质量.对相干斑的抑制是使用SAR数据的必不可少的预处理程序.提出一种基于非局部加权的线性最小均方误差(LMMSE)滤波器的极化SAR滤波的方法.方法 该方法的主要过程是利用非局部均值的理论来获取LMMSE估计器中像素样本的权重.同时,在样本像素的选取过程中,利用待处理像素的极化散射特性和邻域块的异质性来排除不相似像素以加速算法,同时达到保持点目标和自适应调节块窗口大小的目的.结果 模拟影像和真实影像上进行的实验结果表明,采用这种方法滤波后影像的质量得到明显改善.和传统的LMMSE算法相比,无论是单视的影像还是多视的影像,本文方法去噪结果的等效视数都高出8视以上;峰值信噪比也提升了5.8 dB.同时,去噪后影像分类的总体精度也达到了83％以上,该方法的运行效率也比非局部均值算法有了较大提升.结论 本文方法不仅能够有效抑制相干斑噪声,还能较好地保持边缘和细节信息以及极化散射特性.这将会为后续高效利用SAR数据提供保障.     

Imaging and structural feature decomposition of a complex target using multi-aspect polarimetric scattering

Based on the Huynen parametric decomposition of target scattering matrix, the polarimetric ellipse parameters are transformed and applied to decomposition of scattering mechanisms of a complex target in VHR POL-SAR images (very high resolution, polarimetric synthetic aperture radar). Making use of multi-aspect (or circle-aspect) and wideband VHR POL-SAR images, scattering mechanisms of a volumetric target and its structural components are recognized over image pixels. Utilizing the layover features, the target height profile is also estimated from two-dimensional image. As example, polarimetric scattering data of some vehicles on ground, including multi-aspect simulated data and experimental measurements, are applied to validations of scattering mechanism decompositions and target structural feature recognition.     

Crop discrimination based on polarimetric correlation coefficients optimization for PolSAR data

Crop discrimination is a necessary step for most agricultural monitoring systems. Radar polarimetric responses from various crops strongly relate to the types and orientations of the local scatterers, which makes the discrimination still difficult using the polarimetric synthetic aperture radar (PolSAR) technique. This work provides a new approach by investigating and utilizing the characteristics of polarimetric correlation coefficients in the rotation domain along the radar line of sight. The theoretical basis lies in that polarimetric correlation coefficients can reflect the different responses and can be enhanced at different levels for various land-cover types with suitable rotation angles in the rotation domain. In this vein, a polarimetric correlation coefficient optimization framework is established and new polarimetric features are extracted therein. Demonstration with multi-frequency (P-, L-, and C-bands) airborne synthetic aperture radar (AIRSAR) PolSAR data over crop areas validates that polarimetric correlation coefficients are crop dependent and the optimized polarimetric correlation coefficient parameters can better discriminate them. Then, a crop discrimination scheme is proposed using the derived polarimetric features. A flow chart for the optimal discrimination feature set selection and determination is provided and is validated by the real data with seven typical crop types. All these crop types are successfully discriminated for the P- and L-band data, whereas only two types of crops are slightly overlapped in the feature space for the C-band data. Experimental studies demonstrate the efficiency and potential of the established methodology.     

基于相关性分析及遗传算法的高维数据特征选择

特征选择是模式识别及数据挖掘等领域的重要问题之一。针对高维数据对象,特征选择一方面可以提高分类精度和效率,另一方面可以找出富含信息的特征子集。针对此问题,提出了一种综合了filter模型及wrapper模型的特征选择方法,首先基于特征与类别标签的相关性分析进行特征筛选,只保留与类别标签具有较强相关性的特征,然后针对经过筛选而精简的特征子集采用遗传算法进行随机搜索,并采用感知器模型的分类错误率作为评价指标。实验结果表明,该算法可有效地找出具有较好的线性可分离性的特征子集,从而实现降维并提高分类精度。     

全极化SAR影像城区建筑密度分析

针对全极化SAR影像的建筑区特性,提出了一种基于极化特征共生矩阵的城区建筑密度分析方法。首先将极化特征与共生矩阵结合,在考虑建筑区极化散射机理和建筑朝向作用的同时,兼顾了建筑区的空间排列信息,在此基础上为了增强建筑密度的局部区域特性,将共生矩阵特征进行K-means聚类,结合图像分块形成标号直方图统计矢量,进而对该直方图统计矢量进行矢量量化实现SAR影像城区的建筑密度分级。RadarSat-2全极化SAR影像城区建筑密度分析的实验表明,该方法既适用于建筑朝向复杂城区也适用于建筑排列整齐城区的密度信息提取。     

Wrapper–Filter Feature Selection Algorithm Using a Memetic Framework

This correspondence presents a novel hybrid wrapper and filter feature selection algorithm for a classification problem using a memetic framework. It incorporates a filter ranking method in the traditional genetic algorithm to improve classification performance and accelerate the search in identifying the core feature subsets. Particularly, the method adds or deletes a feature from a candidate feature subset based on the univariate feature ranking information. This empirical study on commonly used data sets from the University of California, Irvine repository and microarray data sets shows that the proposed method outperforms existing methods in terms of classification accuracy, number of selected features, and computational efficiency. Furthermore, we investigate several major issues of memetic algorithm (MA) to identify a good balance between local search and genetic search so as to maximize search quality and efficiency in the hybrid filter and wrapper MA     

A PolSAR ship detector based on a multi-polarimetric-feature combination using visual attention

Ship detection can be significantly improved by using polarimetric synthetic aperture radar (PolSAR) imaging. In this article, we propose a PolSAR ship detection method based on the use of multi-featured polarization by using the visual attention model. Three polarimetric features, namely, the polarimetric contrast, the polarimetric scattering, and the polarimetric phase, are selected as the early features, and the pros and cons for each feature are discussed. The visual attention model is a framework that rapidly combines multiple features into one feature, which is improved according to the relationship of the selected features. Validation of the method is performed by analysing the multi-resolution process, the improved multi-feature process, the threshold strategy, the sensibility to the incidence angle of the sensors, and the performance of moving ship detection, which are analysed by Radarsat-2 fine quad images with automatic identification system data. Additionally, the false alarm/non-detection analysis and the computation cost analysis are also considered. In contrast to other ship detectors, the proposed detector is more effective and robust.     

Performance of feature-selection methods in the classification of high-dimension data

Contemporary biological technologies produce extremely high-dimensional data sets from which to design classifiers, with 20,000 or more potential features being common place. In addition, sample sizes tend to be small. In such settings, feature selection is an inevitable part of classifier design. Heretofore, there have been a number of comparative studies for feature selection, but they have either considered settings with much smaller dimensionality than those occurring in current bioinformatics applications or constrained their study to a few real data sets. This study compares some basic feature-selection methods in settings involving thousands of features, using both model-based synthetic data and real data. It defines distribution models involving different numbers of markers (useful features) versus non-markers (useless features) and different kinds of relations among the features. Under this framework, it evaluates the performances of feature-selection algorithms for different distribution models and classifiers. Both classification error and the number of discovered markers are computed. Although the results clearly show that none of the considered feature-selection methods performs best across all scenarios, there are some general trends relative to sample size and relations among the features. For instance, the classifier-independent univariate filter methods have similar trends. Filter methods such as the t-test have better or similar performance with wrapper methods for harder problems. This improved performance is usually accompanied with significant peaking. Wrapper methods have better performance when the sample size is sufficiently large. ReliefF, the classifier-independent multivariate filter method, has worse performance than univariate filter methods in most cases; however, ReliefF-based wrapper methods show performance similar to their t-test-based counterparts.     

Feature selection in single and ensemble learning‐based bankruptcy prediction models

Feature selection is an important data preprocessing step for the construction of an effective bankruptcy prediction model. The prediction performance can be affected by the employed feature selection and classification techniques. However, there have been very few studies of bankruptcy prediction that identify the best combination of feature selection and classification techniques. In this study, two types of feature selection methods, including filter‐ and wrapper‐based methods, are considered, and two types of classification techniques, including statistical and machine learning techniques, are employed in the development of the prediction methods. In addition, bagging and boosting ensemble classifiers are also constructed for comparison. The experimental results based on three related datasets that contain different numbers of input features show that the genetic algorithm as the wrapper‐based feature selection method performs better than the filter‐based one by information gain. It is also shown that the lowest prediction error rates for the three datasets are provided by combining the genetic algorithm with the naïve Bayes and support vector machine classifiers without bagging and boosting.     

改进极化白化滤波的边缘检测

针对SAR影像边缘检测受斑点噪声影响严重和极化信息利用不充分的问题,用滑动模板边缘两侧目标的协方差矩阵代替了极化白化滤波中杂波背景与窗口中心的协方差矩阵,提出一种基于改进极化白化滤波的边缘检测新方法,充分利用了极化通道间的相关性,在有效抑制斑点噪声的同时,提高了极化信息的利用率。模拟和真实极化影像的实验验证了新方法的有效性。     

Wrapper-filter feature selection algorithm using a memetic framework.

This correspondence presents a novel hybrid wrapper and filter feature selection algorithm for a classification problem using a memetic framework. It incorporates a filter ranking method in the traditional genetic algorithm to improve classification performance and accelerate the search in identifying the core feature subsets. Particularly, the method adds or deletes a feature from a candidate feature subset based on the univariate feature ranking information. This empirical study on commonly used data sets from the University of California, Irvine repository and microarray data sets shows that the proposed method outperforms existing methods in terms of classification accuracy, number of selected features, and computational efficiency. Furthermore, we investigate several major issues of memetic algorithm (MA) to identify a good balance between local search and genetic search so as to maximize search quality and efficiency in the hybrid filter and wrapper MA.     

Ordinal-based and frequency-based integration of feature selection methods for sentiment analysis

Feature subset selection with the aim of reducing dependency of feature selection techniques and obtaining a high-quality minimal feature subset from a real-world domain is the main task of this research. For this end, firstly, two types of feature representation are presented for feature sets, namely unigram-based and part-of-speech based feature sets. Secondly, five methods of feature ranking are employed for creating feature vectors. Finally, we propose two methods for the integration feature vectors and feature subsets. An ordinal-based integration of different feature vectors (OIFV) is proposed in order to obtain a new feature vector. The new feature vector depends on the order of features in the old vectors. A frequency-based integration of different feature subsets (FIFS) with most effective features, which are obtained from a hybrid filter and wrapper methods in the feature selection task, is then proposed. In addition, four well-known text classification algorithms are employed as classifiers in the wrapper method for the selection of the feature subsets. A wide range of comparative experiments on five widely-used datasets in sentiment analysis were carried out. The experiments demonstrate that proposed methods can effectively improve the performance of sentiment classification. These results also show that proposed part-of-speech patterns are more effective in their classification accuracy compared to unigram-based features.     

基于目标分解与支持向量机的极化SAR图像分类研究

为了有效地对极化SAR图像进行分类,基于目标分解和支持向量机,提出了一种极化SAR图像非监督分类法。该方法首先利用目标分解理论获得极化熵和平均散射角,并在熵-平均散射角平面对图像进行初分类,以确定类中心;然后利用Wishart分布定义的距离函数寻找训练样本,同时选择一定的极化参数组成特征矢量,并利用训练样本和特征矢量训练支持向量机;最后用训练好的分类器对极化SAR图像进行分类。通过对ESAR图像进行分类,比较了多种参数组合的分类结果,并与Wishart方法进行了比较,结果表明,该方法特征选择非常灵活,不仅结果类内离散度更小,且不需要太多的迭代次数。     

A fast SVM-based wrapper feature selection method driven by a fuzzy complementary criterion

The selection of informative and non-redundant features has become a prominent step in pattern classification. However, despite the intensive research, it is still an open issue to identify valuable feature subsets, especially in highly dimensional feature spaces. This paper proposes a wrapper feature selection method, in the context of support vector machines (SVMs), named Wr-SVM-FuzCoC. Our method combines effectively the advantages of the wrapper and filter approaches, achieving three goals simultaneously: classification performance, dimensionality reduction, and computational efficiency. In the filter part, a forward feature search methodology is developed, driven by a fuzzy complementary criterion, whereby at each iteration a feature is selected that exhibits the maximum additional contribution in regard to the previously selected subset. The quality of single features or feature subsets is assessed via a fuzzy local evaluation criterion with respect to patterns. This is achieved by the so-called fuzzy partition vector (FPV), comprising the fuzzy membership grades of every pattern in their target classes. Derivation of the feature FPVs is accomplished by incorporating a fuzzy output kernel-based support vector machine. The proposed method is favorably compared with existing SVM-based wrapper methods, in terms of performance capability and computational speed. Experimental investigation is carried out using a diverse pool of real datasets, including moderate and high-dimensional feature spaces.