On neural network algorithms for retrieving forest biomass from SAR data

We discuss the application of neural network algorithms (NNAs) for retrieving forest biomass from multifrequency (L- and P-band) multipolarization (hh, vv, and vv) backscattering. After discussing the training and pruning procedures, we examine the performances of neural algorithms in inverting combinations of radar backscattering coefficients at different frequencies and polarization states. The analysis includes an evaluation of the expected sensitivity of the algorithm to measurement noise stemming both from speckle and from fluctuations of vegetation and soil parameters. The NNA accomplishments are compared with those of linear regressions for the same channel combinations. The application of NNAs to invert actual multifrequency multipolarization measurements reported in literature is then considered. The NNA retrieval accuracy is now compared with those yielded by linear and nonlinear regressions and by a model-based technique. A direct analysis of the information content of the radar measurements is finally carried out through an extended pruning procedure of the net.     

Polarimetric radar measurements of a forested area near Mt. Shasta

The authors present the results of an experiment using the NASA/JPL DC-8 AIRSAR (aircraft synthetic-aperture radar) over a coniferous forest near Mt. Shasta (California) in 1989. Calibration devices were deployed in clearings and under the forest canopy and passes at 20°, 40°, and 55° incidence angles were made with the AIRSAR. A total of eight images at differing incidence angles have been processed and calibrated. The multipolarization multifrequency data were examined, and it was found that the C-band cross section averaged over like and cross polarizations is the best parameter for distinguishing between two stands with differing forest biomass. The average cross section at P- and L-bands is useful only for smaller incidence angles. Parameters describing the polarization behavior of the scattering were primarily useful in identifying the dominant scattering mechanisms for forest backscatter     

On the dimensionality of multiparameter microwave image data fromthin sea ice in the Labrador Sea

Thin ice is typically defined as comprising the World Meteorological Organization's young and new ice categories, referring to sea ice that is less than 0.3 m thick. This ice type is an extremely important factor in both the thermodynamic and dynamic properties of both polar and marginal ice covers. In the LIMEX'89 experiment, spatially and temporally registered multipolarization data from 5.3 and 9.25 GHz SARs and from 37 and 90 GHz imaging radiometers were acquired over a region containing a wide range of new ice growth stages in the Labrador ice pack. The temperatures at the data acquisition time were -10 and correspond to ice growth conditions. In this paper, the dimensionality of the multifrequency, multipolarization active and passive data set is examined to determine the complementarity of the sensor parameters and sensor types for thin ice measurements. Principal component analysis is used to provide estimates of the information content of individual measurement channels and their combinations. Various measurement subspaces are examined. Criteria for channel redundance are proposed and tested and the classification potential of the multidimensional measurement set is tested for thin ice growth stages that are known to present classification difficulties for microwave sensors. Given six nominally independent SAR measurement channels, the information space dimension of this subspace is shown to be greater than five. The four radiometer channels are shown to have information space dimension two and are redundant in frequency. The combined, ten element, SAR/radiometer measurement space was shown to have information space dimension eight under the criteria used when the scattering polarization ratios are included     

Application of neural networks for sea ice classification inpolarimetric SAR images

Several automatic methods have been developed to classify sea ice types from fully polarimetric synthetic aperture radar (SAR) images, and these techniques are generally grouped into supervised and unsupervised approaches. In previous work, supervised methods have been shown to yield higher accuracy than unsupervised techniques, but suffer from the need for human interaction to determine classes and training regions. In contrast, unsupervised methods determine classes automatically, but generally show limited ability to accurately divide terrain into natural classes. In this paper, a new classification technique is applied to determine sea ice types in polarimetric and multifrequency SAR images, utilizing an unsupervised neural network to provide automatic classification, and employing an iterative algorithm to improve the performance. The learning vector quantization (LVQ) is first applied to the unsupervised classification of SAR images, and the results are compared with those of a conventional technique, the migrating means method. Results show that LVQ outperforms the migrating means method, but performance is still poor. An iterative algorithm is then applied where the SAR image is reclassified using the maximum likelihood (ML) classifier. It is shown that this algorithm converges, and significantly improves classification accuracy. The new algorithm successfully identifies first-year and multiyear sea ice regions in the images at three frequencies. The results show that L- and P-band images have similar characteristics, while the C-band image is substantially different. Classification based on single features is also carried out using LVQ and the iterative ML method. It is found that the fully polarimetric classification provides a higher accuracy than those based on a single feature. The significance of multilook classification is demonstrated by comparing the results obtained using four-look and single-look classifications     

Observation of sea-ice thickness in the sea of Okhotsk by using dual-frequency and fully polarimetric airborne SAR (pi-SAR) data

To investigate the possibilities of using dual-frequency, multipolarization synthetic aperture radar (SAR) data to monitor sea ice, we derived the relationship between various polarization characteristics and the physical parameters of sea ice. We discuss the frequency and polarization characteristics of the backscattering coefficients of sea ice and then characterize its thickness by comparing the corresponding backscattering coefficient for each polarization with the physical parameters of the ice. We first propose a methodology for classifying sea-ice types by using a polarimetric decomposition technique, before comparing an estimation of the sea-ice thickness with the corresponding dual-frequency, multipolarization SAR data. We utilized the backscattering ratio to estimate the thickness of the sea ice. This ratio canceled the effect of roughness on the backscattering. The method was validated using Pi-SAR (polarimetric and interferometric airborne SAR) observation data obtained at ground-truth sites.     

Spaceborne imaging radar-C instrument

The Shuttle Imaging Radar (SIR)-C instrument has been designed to obtain simultaneous multifrequency and simultaneous multipolarization radar images from a low Earth orbit. It is a multiparameter imaging radar that will be flown during two different seasons. The instrument has been designed to operate in innovative modes such as the squint mode, the extended aperture mode, and the scansar mode, and to demonstrate innovative engineering techniques such as beam nulling for echo tracking, pulse repetition frequency-hopping for Doppler centroid tracking, frequency step chirp generating, for polarization differentiation, and block floating-point quantizing for data compression. The instrument has also been designed to allow flexibility in selection of radar parameters such as pulsewidth and beamwidth in the tradeoff of image quality parameters. These SIR-C capabilities are to be directly transferred to the proposed Earth Observing System (Eos) synthetic aperture radar     

What are the Best Radar Wavelengths, Incidence Angles, and Polanzations for Discrimination among Lava Flows and Sedimentary Rocks? A Statistical Approach

Linear discriminant analysis of multifrequency and multipolarization radar scatterometer data of lava flows and sedimentary rocks indicates that the lava flows sampled at the Snake River Plain, Idaho, and the sedimentary formations sampled at the San Rafael Swell, Utah, can be discriminated from one another. The necessary radar backscatter data, and optimum wavelengths, polarizations, and incidence angles among those available for these problems are indicated by the discriminant analysis program. For separation of the lava flows, shorter wavelengths and smaller incidence angles are best. For the sedimentary rocks, conversely, the longer wavelengths and somewhat larger incidence angles are preferred. Separation among the lava flows was slightly better using horizontally polarized data, while vertical polarization gave better separation with the sedimentary rocks. The method developed here may provide a rationale for user selection tion of operating parameters for advanced radar systems such as the Shuttle Imaging Radar-C (SIR-C) and that proposed for the Earth Observation System (EOS). The analysis is being extended to other study sites for which we have scatterometer data to determine the generality of the results reported here.     

Evaluation of textural and multipolarization radar features forcrop classification

The aim of this research is to evaluate crop discrimination using airborne radar data based on multipolarization and textural information. Multipolarization data (C-HH, C-VV, and C-HV) were used for discriminating 5 crop types i.e., corn, wheat, soya, pasture, and alfalfa. For the multipolarization evaluation, an unsupervised classification algorithm and a supervised method based on maximum likelihood were used on the data. For the textural evaluation, textural measures of different degrees were calculated on three different order histograms and were evaluated from the crop discrimination point of view. Results show that multipolarization correct classification rates of 86.31% and 74.47% were obtained for supervised and unsupervised methods respectively. Hence, multipolarization radar data offer an adequate tool for crop identification especially with supervised classification. The evaluation of textural measures shows that for a first order histogram the mean measure gives the best rate of discrimination. In the case of second and third order histograms, the best measures are contrast and large number emphasis respectively. These textural measures were integrated with the three multipolarization channels in order to determine their specific contributions. Results show that crop class separability is thereby improved and that the rate of correct classification increased by 9.79% for the crops     

A Markov random field model for classification of multisourcesatellite imagery

A general model for multisource classification of remotely sensed data based on Markov random fields (MRF) is proposed. A specific model for fusion of optical images, synthetic aperture radar (SAR) images, and GIS (geographic information systems) ground cover data is presented in detail and tested. The MRF model exploits spatial class dependencies (spatial context) between neighboring pixels in an image, and temporal class dependencies between different images of the same scene. By including the temporal aspect of the data, the proposed model is suitable for detection of class changes between the acquisition dates of different images. The performance of the proposed model is investigated by fusing Landsat TM images, multitemporal ERS-1 SAR images, and GIS ground-cover maps for land-use classification, and on agricultural crop classification based on Landsat TM images, multipolarization SAR images, and GIS crop field border maps. The performance of the MRF model is compared to a simpler reference fusion model. On an average, the MRF model results in slightly higher (2%) classification accuracy when the same data is used as input to the two models. When GIS field border data is included in the MRF model, the classification accuracy of the MRF model improves by 8%. For change detection in agricultural areas, 75% of the actual class changes are detected by the MRF model, compared to 62% for the reference model. Based on the well-founded theoretical basis of Markov random field models for classification tasks and the encouraging experimental results in our small-scale study, the authors conclude that the proposed MRF model is useful for classification of multisource satellite imagery     

Classification of Baltic Sea ice types by airborne multifrequencymicrowave radiometer

An airborne multifrequency radiometer (24, 34, 48, and 94 GHz, vertical polarization) was used to investigate the behavior of the brightness temperature of different sea ice types in the Gulf of Bothnia (Baltic Sea). The measurements and the main results of the analysis are presented. The measurements were made in dry and wet conditions (air temperature above and below 0°C). The angle of incidence was 45° in all measurements. The following topics are evaluated: a) frequency dependency of the brightness temperature of different ice types, b) the capability of the multifrequency radiometer to classify ice types for winter navigation purposes, and c) the optimum measurement frequencies for mapping sea ice. The weather conditions had a significant impact on the radiometric signatures of some ice types (snow-covered compact pack ice and frost-covered new ice); the impact was the highest at 94 GHz. In all cases the overall classification accuracy was around 90% (the kappa coefficient was from 0.86 to 0.96) when the optimum channel combination (24/34 GHz and 94 GHz) was used     

Delineation of inundated area and vegetation along the Amazonfloodplain with the SIR-C synthetic aperture radar

Floodplain inundation and vegetation along the Negro and Amazon rivers near Manaus, Brazil were accurately delineated using multi-frequency, polarimetric synthetic aperture radar (SAR) data from the April and October 1994 SIR-C missions. A decision-tree model was used to formulate rules for a supervised classification into five categories: water, clearing (pasture), aquatic macrophyte (floating meadow), nonflooded forest, and flooded forest. Classified images were produced and tested within three days of SIR-C data acquisition. Both C-band (5.7 cm) and L-band (24 cm) wavelengths were necessary to distinguish the cover types. HH polarization was most useful for distinguishing flooded from nonflooded vegetation (C-HH for macrophyte versus pasture, and L-HH for flooded versus nonflooded forest), and cross-polarized L-band data provided the best separation between woody and nonwoody vegetation. Between the April and October missions, the Amazon River level fell about 3.6 m and the portion of the study area covered by flooded forest decreased from 23% to 12%. This study demonstrates the ability of multifrequency SAR to quantify in near realtime the extent of inundation on forested floodplains, and its potential application for timely monitoring of flood events     

Multipolarization SAR Data for Surface Feature Delineation and Forest Vegetation Characterization

This paper presents the utility of multipolarization Synthetic Aperture Radar (SAR) data for surface feature delineation and forest vegetation characterization. Three channels of ratioed data (VV/HH, VH/HH, and VH/VV) are generated from the HH, VV, and VH polarization data (V = vertical, H = horizontal). The ratioed data are linearly stretched to yield a digital number within a range of 0 to 255. The techniques for reducing SAR speckle noise and for measuring the degree of separation are discussed. For surface feature delineation, the results indicate that cross polarization as well as cross polarization ratioed data better delineate those surface features that are difficult to separate by like polarization data. The results suggest using a median value filtering technique to reduce within-plot data fluctuation to increase the separability measure. For forest vegetation characterization, the results indicate that multipolarization SAR data may be used to estimate forest properties such as total-tree biomass, basal area, and tree height.     

结合ICA相干斑抑制的全极化SAR图像分类

本文提出了一种结合相干斑抑制的全极化 SAR(Synthetic Aperture Radar)图像分类新方法.该方法先对图像数据做Pauli分解,获得三个极化组合通道,并分别用三种颜色表示这三个极化组合;再用独立分量分析稀疏编码(ICA-SCS)算法对各颜色通道进行相干斑抑制,最后把三个颜色通道混合,实现了对图像信息的分类.该方法很好的保留了极化通道间的相对相位信息,同时,相干斑抑制后的数据直接用于图像分类,不需要再做任何极化通道组合.对真实SAR图像的分类结果表明,该方法对分类效果和精度有明显改善.     

Potential Application of Multipolarization SAR for Pine-Plantation Biomass Estimation

This paper presents the techniques and the potential utility of multipolarization Synthetic Aperture Radar (SAR) data for pineplantation biomass estimation. Three channels of SAR data, one from the Shuttle Imaging Radar SIR-A and the other two from the aircraft SAR, were acquired over the Baldwin County, Alabama, study area. The SIR-A data were acquired with HH polarization and the aircraft SAR data with VV and VH polarizations. Linear regression techniques are used to estimate the pine-plantation biomass, tree height, and age using 21 test plots. The results indicate that the multipolarization data are highly related to the plantation biomass. The results suggest a potential application of multipolarization SAR for pine-plantation biomass estimation.     

Radar imaging of moving targets in foliage using multifrequency multiaperture polarimetric SAR

Because of the low signal-to-clutter ratio, it is a difficult problem to detect and image moving targets in foliage. In this paper, a multifrequency multiaperture polarimetric synthetic aperture radar (MFMA POLSAR) system is proposed for imaging of moving targets in foliage. The MFMA POLSAR extends the multifrequency antenna array SAR (MF-SAR) system to multiple polarizations. Full polarization is used in MFMA POLSAR to achieve an optimal polarization adaptive to the environment such that the images obtained by different apertures are of the best coherence that is used to obtain the highest accuracy of the phase estimation. It is also shown that the MFMA POLSAR cannot only accurately locate both the slow and the fast moving targets but also reveal moving targets in foliage.     

一种基于集成学习和特征融合的遥感影像分类新方法

针对多源遥感数据分类的需要,提出了一种基于全极化SAR影像、极化相干矩阵特征、光学遥感影像光谱和纹理的多种特征融合和多分类器集成的遥感影像分类新方法.对全极化PALSAR数据进行预处理和极化相干矩阵特征提取,利用灰度共生矩阵计算光学和SAR影像的对比度、逆差距、二阶距、差异性等纹理特征参数,并与光谱特征结合,形成6种组合策略.利用集成学习方法对随机森林分类器、子空间分类器、最小距离分类器、支持向量机分类器、反向传播神经网络分类器等分类器进行组合,对不同组合策略的遥感影像特征集进行分类.结果表明提出的基于多种特征和多分类器集成的新方法很好地利用了主被动遥感数据在不同地表景观类型提取上的潜力,综合了多种算法的优势,能够有效地提高总体精度和各类别的分类精度.     

高分辨率多极化SAR图像目标分类误差分析

讨论了多极化SAR系统的极化误差及其对多极化SAR图像极化匹配目标分类性能的影响，并给出了计算结果。分析表明，多极化SAR系统的极化通道幅度不平衡误差对目标极化匹配结果的影响最大，交叉极化干扰项对此也有较大的影响。     

Adaptive segmentation of MRI data

Intensity-based classification of MR images has proven problematic, even when advanced techniques are used. Intrascan and interscan intensity inhomogeneities are a common source of difficulty. While reported methods have had some success in correcting intrascan inhomogeneities, such methods require supervision for the individual scan. This paper describes a new method called adaptive segmentation that uses knowledge of tissue intensity properties and intensity inhomogeneities to correct and segment MR images. Use of the expectation-maximization (EM) algorithm leads to a method that allows for more accurate segmentation of tissue types as well as better visualization of magnetic resonance imaging (MRI) data, that has proven to be effective in a study that includes more than 1000 brain scans. Implementation and results are described for segmenting the brain in the following types of images: axial (dual-echo spin-echo), coronal [three dimensional Fourier transform (3-DFT) gradient-echo T1-weighted] all using a conventional head coil, and a sagittal section acquired using a surface coil. The accuracy of adaptive segmentation was found to be comparable with manual segmentation, and closer to manual segmentation than supervised multivariant classification while segmenting gray and white matter.     

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 combination 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 classification accuracy, fast convergence and high stability. The effectiveness of this algorithm is demonstrated by experiments using SIR-C/X-SAR （Spaceborne Imaging Radar-C/X-band Synthetic Aperture Radar） data.     

Electrical impedance spectroscopy of the breast: clinical imaging results in 26 subjects

Electrical impedance spectroscopy (EIS) is a potential, noninvasive technique to image women for breast cancer. Studies have shown characteristic frequency dispersions in the electrical conductivity and permittivity of malignant versus normal tissue. Using a multifrequency EIS system, we imaged the breasts of 26 women. All patients had mammograms ranked using the American College of Radiology (ACR) BIRADS system. Of the 51 individual breasts imaged, 38 were ACR 1 negative, six had ACR 4-5 suspicious lesions, and seven had ACR 2 benign findings such as fibroadenomas or calcifications. A radially translatable circular array of 16 Ag/AgCl electrodes was placed around the breast while the patient lay prone. We applied trigonometric voltage patterns at ten frequencies between 10 and 950 kHz. Anatomically coronal images were reconstructed from this data using nonlinear partial differential equation methods. Typically, ACR 1-rated breasts were interrogated in a single central plane whereas ACR 2-5-rated breasts were imaged in multiple planes covering the region of suspicion. In general, a characteristic homogeneous image emerged for mammographically normal cases while focal inhomogeneities were observed in images from women with malignancies. Using a specific visual criterion, EIS images identified 83% of the ACR 4-5 lesions while 67% were detected using a numerical criterion. Overall, multifrequency electrical impedance imaging appears promising for detecting breast malignancies, but improvements must be made before the method reaches its full potential.