极化SAR图像分割方法研究

极化SAR图像分割是面向对象的极化SAR图像分析处理的重要组成部分,也是极化SAR图像处理的关键和难题。然而,目前还没有一种极化SAR分割方法被广泛接受。文章通过对现有的极化SAR图像分割方法进行综述,以使各位研究者对其有一个较全面的认识。文章首先介绍了国内外在极化SAR图像分割方面的主要研究机构;然后对现有的极化SAR图像分割算法进行了分类,并归纳了不同方法的基本思想,分析了各自的性能特点;最后对极化SAR图像分割方法的研究现状及发展趋势进行了总结和展望。     

基于复Wishart分布的极化SAR图像无监督分类

文中使用了一种无监督算法对全极化合成孔径雷达数据进行地物分类。不同于其他算法对像素统计特性分类而忽略他们的散射特性,这种算法不仅使用了统计分类,而且还保留了其主要的散射特性。本算法采用的是由Freeman和Durden以散射模型为基础开发的分解算法和基于复Wishart分布的距离度量分类器的组合。首先是应用Freeman和Durden分解划分像素分成三个散射类:表面散射,体散射和二面角散射;然后在此基础上将这三个散射类分为多个聚类,通过迭代Wishart分类器将其更精准地分类;最后,根据散射类别的自然颜色对其进行像素编码,提出颜色填充的方案。通过实验结果比对来证明该方法比H/α方法更具有优越性。     

Measurement of topography using polarimetric SAR images

A processing technique for polarimetric synthetic aperture radar (SAR) data has been developed which produces profiles of terrain slopes and elevations in the azimuthal (or along-track) direction. This technique estimates the average shift in orientation angle of copolarization backscatter caused by azimuthal tilts of the scattering plane. Using P-band data, tests of this technique have been made for an area in the Black Forest near Villingen/Schwenningen in Baden-Wurttemberg, Germany. The radar measured slope and derived elevation profiles have low rms errors and high correlation values when compared with a stereo-photograph digital-elevation map (DEM) for the area. This algorithm is capable of adaptively making transitions from the forested areas to nearby regions with open-terrain. Subsequent tests of the algorithm have been conducted using polarimetric SAR L-band data for a mountainous, nonforested, region in the Mojave Desert (Ft. Irwin, CA) where an accurate DEM also was available. Complete elevation and slope mapping of the terrain in two dimensions using this technique is possible when azimuthal elevation profiles are produced throughout the range extent of the SAR image     

A three-component scattering model for polarimetric SAR data

An approach has been developed that involves the fit of a combination of three simple scattering mechanisms to polarimetric SAR observations. The mechanisms are canopy scatter from a cloud of randomly oriented dipoles, evenor double-bounce scatter from a pair of orthogonal surfaces with different dielectric constants and Bragg scatter from a moderately rough surface. This composite scattering model is used to describe the polarimetric backscatter from naturally occurring scatterers. The model is shown to describe the behavior of polarimetric backscatter from tropical rain forests quite well by applying it to data from NASA/Jet Propulsion Laboratory's (JPLs) airborne polarimetric synthetic aperture radar (AIRSAR) system. The model fit allows clear discrimination between flooded and nonflooded forest and between forested and deforested areas, for example. The model is also shown to be usable as a predictive tool to estimate the effects of forest inundation and disturbance on the fully polarimetric radar signature. An advantage of this model fit approach is that the scattering contributions from the three basic scattering mechanisms can be estimated for clusters of pixels in polarimetric SAR images. Furthermore, it is shown that the contributions of the three scattering mechanisms to the HH, HV, and VV backscatter can be calculated from the model fit. Finally, this model fit approach is justified as a simplification of more complicated scattering models, which require many inputs to solve the forward scattering problem     

Restoration of polarimetric SAR images using simulated annealing

Filtering synthetic aperture radar (SAR) images ideally results in better estimates of the parameters characterizing the distributed targets in the images while preserving the structures of the nondistributed targets. However, these objectives are normally conflicting, often leading to a filtering approach favoring one of the objectives. An algorithm for estimating the radar cross-section (RCS) for intensity SAR images has previously been proposed in the literature based on Markov random fields and the stochastic optimization method simulated annealing. A new version of the algorithm is presented applicable to multilook polarimetric SAR images, resulting in an estimate of the mean covariance matrix rather than the RCS. Small windows are applied in the filtering, and due to the iterative nature of the approach, reasonable estimates of the polarimetric quantities characterizing the distributed targets are obtained while at the same time preserving most of the structures in the image. The algorithm is evaluated using multilook polarimetric L-band data from the Danish airborne EMISAR system, and the impact of the algorithm on the unsupervised H-α classification is demonstrated     

Unsupervised regions based segmentation using object discovery

We present a new unsupervised algorithm to discovery and segment out common objects from multiple images. Compared with previous cosegmentation methods, our algorithm performs well even when the appearance variations in the foregrounds are more substantial than those in some areas of the backgrounds. Our algorithm mainly includes two parts: the foreground object discovery scheme and the iterative region allocation algorithm. Two terms, a region-saliency prior and a region-repeatness measure, are introduced in the foreground object discovery scheme to detect the foregrounds without any supervisory information. The iterative region allocation algorithm searches the optimal solution for the final segmentation with the constraints from a maximal spanning tree, and an effective color-based model is utilized during this process. The comparative experimental results show that the proposed algorithm matches or outperforms several previous methods on several standard datasets.     

Unsupervised Bayesian image segmentation using orthogonal series

This paper deals with the problem of unsupervised image segmentation which consists in first mixture identification phase and second a Bayesian decision phase. During the mixture identification phase, the conditional probability density function (pdf) and the a priori class probabilities must be estimated. The most difficult part is the estimation of the number of pixel classes or in other words the estimation of the number of density mixture components. To resolve this problem, we propose here a Stochastic and Nonparametric Expectation-Maximization (SNEM) algorithm. The algorithm finds the most likely number of classes, their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. The non-parametric aspect comes from the use of the orthogonal series estimator. Experimental results are promising, we have obtained accurate results on a variety of real images.     

Single-baseline polarimetric SAR interferometry

Examines the application of single-baseline polarimetric SAR interferometry to the remote sensing and measurement of structure over forested terrain. For this, a polarimetric coherent scattering model for vegetation cover suitable for the estimation of forest parameters from interferometric observables is introduced, discussed and validated. Based on this model, an inversion algorithm which allows the estimation of forest parameters such as tree height, average extinction, and underlying topography from single-baseline fully polarimetric interferometric data is addressed. The performance of the inversion algorithm is demonstrated using fully polarimetric single baseline experimental data acquired by DLR's E-SAR system at L-band     

Calibration of linearly polarized polarimetric SAR data subject to Faraday rotation

A model for linearly polarized fully polarimetric backscatter measurements is used, incorporating the effects of system noise, channel amplitude, phase imbalance, crosstalk, and Faraday rotation. A step-by-step procedure is outlined for correction (or calibration) of fully polarimetric data subject to Faraday rotation, to recover the true scattering matrix. The procedure identifies steps for crosstalk removal and correction of channel imbalances that are robust in the presence of Faraday rotation. The final steps in the procedure involve a novel strategy for estimation and correction of Faraday rotation. Three approaches to estimate the (one-way) Faraday rotation angle /spl Omega/ directly from linear (quad-) polarized synthetic aperture radar (SAR) backscatter data obtained by a spaceborne SAR system are described. Each approach can initially be applied to the signature of any scatterer within the scene. Sensitivity analyses are presented that show that at least one of the measures can be used to estimate /spl Omega/ to within /spl plusmn/3/spl deg/ to 5/spl deg/, with reasonable levels of residual crosstalk, noise floor, channel amplitude, and phase imbalance. Ambiguities may be present in the estimates of /spl Omega/ of /spl plusmn/n/spl pi//2 - the impact of this is discussed, and several approaches are suggested to deal with this possibility. The approach described in this paper is relevant for future L-band spaceborne SARs and removes one key obstacle to the deployment of even longer wavelength SARs (e.g., an ultrahigh frequency or P-band SAR) in Earth orbit.     

Knowledge-based segmentation of SAR data with learned priors

An approach for the segmentation of still and video synthetic aperture radar (SAR) images is described. A priori knowledge about the objects present in the image, e.g., target, shadow and background terrain, is introduced via Bayes' rule. Posterior probabilities obtained in this way are then anisotropically smoothed, and the image segmentation is obtained via MAP classifications of the smoothed data. When segmenting sequences of images, the smoothed posterior probabilities of past frames are used to learn the prior distributions in the succeeding frame. We show with examples from public data sets that this method provides an efficient and fast technique for addressing the segmentation of SAR data.     

Unsupervised segmentation of textured images using a hierarchicalneural structure

A hierarchical learning structure, combining a randomly-placed local window, a self-organising map and a local-voting scheme, has been developed for the unsupervised segmentation of textured images, which are modelled by Markov random fields. The system learns to progressively estimate model parameters, and hence classify the various textured regions. A globally correct segregation has consistently been obtained during extensive experiments on both synthetic and natural textured images     

Unsupervised texture segmentation of images using tuned matchedGabor filters

Recent studies have confirmed that the multichannel Gabor decomposition represents an excellent tool for image segmentation and boundary detection. Unfortunately, this approach when used for unsupervised image analysis tasks imposes excessive storage requirements due to the nonorthogonality of the basis functions and is computationally highly demanding. In this correspondence, we propose a novel method for efficient image analysis that uses tuned matched Gabor filters. The algorithmic determination of the parameters of the Gabor filters is based on the analysis of spectral feature contrasts obtained from iterative computation of pyramidal Gabor transforms with progressive dyadic decrease of elementary cell sizes. The method requires no a priori knowledge of the analyzed image so that the analysis is unsupervised. Computer simulations applied to different classes of textures illustrate the matching property of the tuned Gabor filters derived using our determination algorithm. Also, their capability to extract significant image information and thus enable an easy and efficient low-level image analysis will be demonstrated.     

Unsupervised terrain classification preserving polarimetric scattering characteristics

In this paper, we proposed an unsupervised terrain and land-use classification algorithm using polarimetric synthetic aperture radar data. Unlike other algorithms that classify pixels statistically and ignore their scattering characteristics, this algorithm not only uses a statistical classifier, but also preserves the purity of dominant polarimetric scattering properties. This algorithm uses a combination of a scattering model-based decomposition developed by Freeman and Durden and the maximum-likelihood classifier based on the complex Wishart distribution. The first step is to apply the Freeman and Durden decomposition to divide pixels into three scattering categories: surface scattering, volume scattering, and double-bounce scattering. To preserve the purity of scattering characteristics, pixels in a scattering category are restricted to be classified with other pixels in the same scattering category. An efficient and effective class initialization scheme is also devised to initially merge clusters from many small clusters in each scattering category by applying a merge criterion developed based on the Wishart distance measure. Then, the iterative Wishart classifier is applied. The stability in convergence is much superior to that of the previous algorithm using the entropy/anisotropy/Wishart classifier. Finally, an automated color rendering scheme is proposed, based on the classes' scattering category to code the pixels to resemble their natural color. This algorithm is also flexible and computationally efficient. The effectiveness of this algorithm is demonstrated using the Jet Propulsion Laboratory's AIRSAR and the German Aerospace Center's (DLR) E-SAR L-band polarimetric synthetic aperture radar images.     

Speckle filtering in polarimetric SAR data based on the subspace decomposition

In this paper, a new approach to speckle filtering of synthetic aperture radar (SAR) data is presented. We define a parameter space consisting of two orthogonal subspaces-the signal subspace and the noise subspace. Then, the full polarimetric information from the signal subspace is obtained after speckle filtering. In this way, edges of different kinds of targets are preserved. The effectiveness of this method is demonstrated using the National Aeronautics and Space Administration Jet Propulsion Laboratory airborne L-band polarimetric SAR data.     

Unsupervised texture segmentation using multichannel decompositionand hidden Markov models

In this paper, we describe an automatic unsupervised texture segmentation scheme using hidden Markov models (HMMs). First, the feature map of the image is formed using Laws' micromasks and directional macromasks. Each pixel in the feature map is represented by a sequence of 4-D feature vectors. The feature sequences belonging to the same texture are modeled as an HMM. Thus, if there are M different textures present in an image, there are M distinct HMMs to be found and trained. Consequently, the unsupervised texture segmentation problem becomes an HMM-based problem, where the appropriate number of HMMs, the associated model parameters, and the discrimination among the HMMs become the foci of our scheme. A two-stage segmentation procedure is used. First, coarse segmentation is used to obtain the approximate number of HMMs and their associated model parameters. Then, fine segmentation is used to accurately estimate the number of HMMs and the model parameters. In these two stages, the critical task of merging the similar HMMs is accomplished by comparing the discrimination information (DI) between the two HMMs against a threshold computed from the distribution of all DI's. A postprocessing stage of multiscale majority filtering is used to further enhance the segmented result. The proposed scheme is highly suitable for pipeline/parallel implementation. Detailed experimental results are reported. These results indicate that the present scheme compares favorably with respect to other successful schemes reported in the literature.     

Knowledge-based classification of polarimetric SAR images

In preparation for the flight of the Shuttle Imaging Radar-C (SIR-C) on board the Space Shuttle in the spring of 1994, a level-1 automatic classifier was developed on the basis of polarimetric SAR images acquired by the JPL AirSAR system. The classifier uses L- and C-Band polarimetric SAR measurements of the imaged scene to classify individual pixels into one of four categories: tall vegetation (trees), short vegetation, urban, or bare surface, with the last category encompassing water surfaces, bare soil surfaces, and concrete or asphalt-covered surfaces. The classifier design uses knowledge of the nature of radar backscattering from surfaces and volumes to construct appropriate discriminators in a sequential format. The classifier, which was developed using training areas in a test site in Northern Michigan, was tested against independent test areas in the same test site and in another site imaged three months earlier. Among all cases and all categories, the classification accuracy ranged between 91% and 100%     

Calibration of a polarimetric imaging SAR

Calibration using point targets is discussed. The fourport network calibration technique is used to describe the radar error model. The processor ambiguity function and the radar distortion matrices are combined to form a generalized polarimetric ambiguity function. The polarimetric ambiguity function of the SAR is found using a single point target, namely a trihedral corner reflector. Based on the resultant polarimetric ambiguity function, an estimate for the backscattering coefficient of the terrain is found using a modified version of the single target calibration technique (STCT). A radar image recorded by the JPL aircraft SAR, which includes a variety of point targets, is used for verification of the new calibration method. The calibrated responses of the point targets are compared both with theory and responses based on the POLCAL technique     

Classification of multifrequency polarimetric SAR imagery using adynamic learning neural network

A practical method for extracting microwave backscatter for terrain-cover classification is presented. The test data are multifrequency (P, L, C bands) polarimetric SAR data acquired by JPL over an agricultural area called “Flevoland”. The terrain covers include forest, water, bare soil, grass, and eight other types of crops. The radar response of crop types to frequency and polarization states were analyzed for classification based on three configurations: 1) multifrequency and single-polarization images; 2) single-frequency and multipolarization images; and 3) multifrequency and multipolarization images. A recently developed dynamic learning neural network was adopted as the classifier. Results show that using partial information, P-band multipolarization images and multiband hh polarization images have better classification accuracy, while with a full configuration, namely, multiband and multipolarization, gives the best discrimination capability. The overall accuracy using the proposed method can be as high as 95% with a total of thirteen cover types classified. Further reduction of the data volume by means of correlation analysis was conducted to single out the minimum data channels required. It was found that this method efficiently reduces the data volume while retaining highly acceptable classification accuracy     

Unsupervised foreground segmentation using background elimination and graph cut techniques

A simple and unsupervised approach to segmenting foreground regions is proposed. This is a novel method for extracting foreground regions from still images by background elimination and graph cut techniques. To extract foreground regions effectively, a new method of background elimination is proposed to detect candidate object regions and a graph cut is used to extract exact foregrounds from the candidate object regions. Experimental results have shown that the proposed method achieves better performance of foreground extraction than existing methods under various environments containing multiple objects and clutter backgrounds in natural images.     

Unsupervised classification of polarimetric synthetic aperture Radar images using fuzzy clustering and EM clustering

Five clustering techniques are compared by classifying a polarimetric synthetic aperture radar image. The pixels are complex covariance matrices, which are known to have the complex Wishart distribution. Two techniques are fuzzy clustering algorithms based on the standard /spl lscr//sub 1/ and /spl lscr//sub 2/ metrics. Two others are new, combining a robust fuzzy C-means clustering technique with a distance measure based on the Wishart distribution. The fifth clustering technique is an application of the expectation-maximization algorithm assuming the data are Wishart. The clustering algorithms that are based on the Wishart are demonstrably more effective than the clustering algorithms that appeal only to the /spl lscr//sub p/ norms. The results support the conclusion that the pixel model is more important than the clustering mechanism.