This PDF file contains the editorial “Special Section Guest Editorial: Sparsity Driven High Dimensional Remote Sensing Image Processing and Analysis” for JRS Vol. 10 Issue 04

Water hyacinth plants (Eichhornia crassipes) are threatening freshwater ecosystems throughout Africa. The Neochetina spp. weevils are seen as an effective solution that can combat the proliferation of the invasive alien plant. We aimed to determine if multitemporal hyperspectral data could be utilized to detect the efficacy of the biocontrol agent. The random forest (RF) algorithm was used to classify variable infestation levels for 6 weeks using: (1) all the hyperspectral bands, (2) bands selected by the recursive feature elimination (RFE) algorithm, and (3) bands selected by the Boruta algorithm. Results showed that the RF model using all the bands successfully produced low-classification errors (12.50% to 32.29%) for all 6 weeks. However, the RF model using Boruta selected bands produced lower classification errors (8.33% to 15.62%) than the RF model using all the bands or bands selected by the RFE algorithm (11.25% to 21.25%) for all 6 weeks, highlighting the utility of Boruta as an all relevant band selection algorithm. All relevant bands selected by Boruta included: 352, 754, 770, 771, 775, 781, 782, 783, 786, and 789 nm. It was concluded that RF coupled with Boruta band-selection algorithm can be utilized to undertake multitemporal monitoring of variable infestation levels on water hyacinth plants.

The thermal airborne hyperspectral imager (TASI), which has 32 channels that provide continuous spectral coverage within wavelengths of 8 to 11.5  μm, is very beneficial for land surface temperature and land surface emissivity (LSE) retrieval. In remote sensing applications, emissivity is important for features classification and temperature is important for environmental monitoring, global climate change, and target recognition studies. This paper proposed a temperature and emissivity separation method via sparse representation (SR-TES) with TASI data, which employs a sparseness differences point of view whereby the atmospheric spectrum cannot be considered SR under the LSE spectral dictionary. We built the dictionary from Johns Hopkins University’s spectral library as an overcomplete base, and the dictionary learning K-SVD algorithm was adopted. The simulation results showed that SR-TES performed better than the TES algorithm in the case of noise impact, and the results from TASI data for the Liuyuan research region were reasonable; partial validation revealed a root mean square error of 0.0144 for broad emissivity, which preliminarily proves that this method is feasible.

This paper proposes to combine locality-preserving projections (LPP) and sparse representation (SR) for hyperspectral image classification. The LPP is first used to reduce the dimensionality of all the training and testing data by finding the optimal linear approximations to the eigenfunctions of the Laplace Beltrami operator on the manifold, where the high-dimensional data lies. Then, SR codes the projected testing pixels as sparse linear combinations of all the training samples to classify the testing pixels by evaluating which class leads to the minimum approximation error. The integration of LPP and SR represents an innovative contribution to the literature. The proposed approach, called locality-preserving SR-based classification, addresses the imbalance between high dimensionality of hyperspectral data and the limited number of training samples. Experimental results on three real hyperspectral data sets demonstrate that the proposed approach outperforms the original counterpart, i.e., SR-based classification.

High-resolution remote-sensing images are increasingly applied in land-use classification problems. Land-use scenes are often very complex and difficult to represent. Subsequently, the recognition of multiple land-cover classes is a continuing research question. We propose a classification framework based on a sparse coding-based correlaton (termed sparse correlaton) model to solve this challenge. Specifically, a general mapping strategy is presented to label visual words and generate sparse coding-based correlograms, which can exploit the spatial co-occurrences of visual words. A compact spatial representation without loss discrimination is achieved through adaptive vector quantization of correlogram in land-use scene classification. Moreover, instead of using K-means for visual word encoding in the original correlaton model, our proposed sparse correlaton model uses sparse coding to achieve lower reconstruction error. Experiments on a public ground truth image dataset of 21 land-use classes demonstrate that our sparse coding-based correlaton method can improve the performance of land-use scene classification and outperform many existing bag-of-visual-words-based methods.

A one-class sparse representation classifier (OCSRC) is proposed to solve the multitemporal change detection problem for identifying disaster affected areas. The OCSRC method, which is adapted from a sparse representation classifier (SRC), incorporates the one-class strategy from a one-class support vector machine (OCSVM) to seek accurate representation for the class of changed areas. It assumes that pixels from the changed areas can be well represented by samples from this class, thus the representation errors are taken as the possibilities of change. Performances of OCSRC and OCSVM are tested and compared with multitemporal multispectral HJ-1A images acquired in Heilongjiang Province before and after the flood in 2013. The entire image, together with two subimages, are used for overall comparison and detailed discussion. Receiver-operating-characteristics curve results show that OCSRC outperforms OCSVM by a lower false-positive rate at a defined true-positive rate (TPR), and the gap is more obvious with high TPR values. The same outcome is also manifested in the change detection image results, with less misclassified pixels for OCSRC at certain TPR values, which implies a more accurate description of the changed area.

Reducing the size of the data on-ground with no information loss represents a strong challenge for the scientific community, since Earth observation (EO) data volumes have strongly and steadily grown during the last 10 years and the need for more efficient compression methods is growing stronger. High-accuracy processing methods employed for EO data understanding and quantifying may result in effective methods for image compression. We propose to use a robust framework of endmember extraction and nonlinear modeling for the on-ground compression of EO data records, where the distribution of the mixture coefficient is exploited to enhance the compression gain while providing high-accuracy reconstruction. Experimental results over real EO datasets show the actual power of the proposed approach.

Segmentation of real-world remote sensing images is a challenge due to the complex texture information with high heterogeneity. Thus, graph-based image segmentation methods have been attracting great attention in the field of remote sensing. However, most of the traditional graph-based approaches fail to capture the intrinsic structure of the feature space and are sensitive to noises. A ℓ_1/2-norm regularization-based graph segmentation method is proposed to segment remote sensing images. First, we use the occlusion of the random texture model (ORTM) to extract the local histogram features. Then, a ℓ_1/2-norm regularized low-rank and sparse representation (L_1/2NNLRS) is implemented to construct a ℓ_1/2-regularized nonnegative low-rank and sparse graph (L_1/2NNLRS-graph), by the union of feature subspaces. Moreover, the L_1/2NNLRS-graph has a high ability to discriminate the manifold intrinsic structure of highly homogeneous texture information. Meanwhile, the L_1/2NNLRS representation takes advantage of the low-rank and sparse characteristics to remove the noises and corrupted data. Last, we introduce the L_1/2NNLRS-graph into the graph regularization nonnegative matrix factorization to enhance the segmentation accuracy. The experimental results using remote sensing images show that when compared to five state-of-the-art image segmentation methods, the proposed method achieves more accurate segmentation results.

Spectral unmixing aims at finding the spectrally pure constituent materials (also called endmembers) and their respective fractional abundances in each pixel of the hyperspectral image scene. One important issue in hyperspectral data unmixing is the initialization of endmembers. Most unmixing methods initialize their endmembers by randomly selecting a specified number of pixels from the data or by vertex component analysis, which limits their performance in practice. We propose an endmember initialization method for hyperspectral data unmixing. Our initial endmembers include some of the true endmembers, which improves the accuracy of hyperpspectral unmixing effectively. The experimental results on both synthetic and real hyperspectral data illustrate the superiority of the proposed method compared with other state-of-the-art approaches.

The imagery vendors of the most advanced remote sensing satellites usually only provide the coefficients of rational function model (RFM) to replace the sensor model and the precise imaging parameters (orbit parameter, attitude parameter, and so on). So, the rigorous imaging model was limited to use in the geometric correction of remote sensing image. The RFM method could obtain a better correction performance in most cases. However, when the image contains few numbers or uneven distribution of ground control points (GCPs), such as infrared image, the RFM method could not obtain the expected performance. Therefore, a geometric correction method for linear pushbroom infrared imagery using compressive sampling (CS) is proposed. The core idea of the proposed method is to use the equivalent bias angles to approximate the influence of the errors (thermal distortion, optical distortion, assembly error, satellite orbit errors, attitude errors, and so on) in the imaging process and adopt the CS method to recover the equivalent bias angle signals. Most of the data are processed scene by scene with enough GCPs for each scene in conventional methods. This restriction is broken by using the sparsity of equivalent bias angle signals in the proposed method. The infrared images from the Hyperion of EO-1 are used as experiment data, and the results of experiments demonstrate the feasibility and superior performance of proposed method.

Hyperspectral remote sensing provides the possibility of direct detection of material information; however, coarse spatial resolution can restrict the scope of its application. The super-resolution (SR) technique can overcome this problem, but the separate application of SR reconstruction to each spectral band is computationally intensive. We proposed an approach that combines empirical mode decomposition (EMD), single-image SR reconstruction using compressed sensing (CS), and principal component analysis (PCA). EMD was used to extract details from within the images, whereas PCA was implemented to reduce the spectral dimensions of the hyperspectral image cube and to retain meaningful spectral information. The CS-based single-image SR reconstruction involved the use of both the K-SVD algorithm for learning and obtaining an over-complete dictionary, and the orthogonal matching pursuit algorithm for the image reconstruction. Experimental results obtained using an EO-1 hyperion image were used to validate the proposed approach.

A mobile light detecting and ranging (LiDAR) system is used to provide point cloud datasets as a topographic base for runoff studies. The point clouds are rasterized to evaluate road runoff using the D8 algorithm. Gaussian noise is artificially induced in the point cloud to simulate inaccuracies in geopositioning and determine its influence in the evaluation of runoff direction. Accuracy in the determination of flow direction decreases with the increase of Gaussian noise. Accuracy also decreases with the decrease of the cell size of the raster dataset. Flow direction shows inaccuracies up to 47 deg with a cell resolution of 0.5 m and Gaussian noise of 0.15 m (standard deviation). On the other hand, cell resolutions of 5 m show a maximum difference of 15 deg with the same noise.

We present a stripmap-mode raw data generator that accounts for trajectory deviations with a nonzero squint angle for an extended scene. The approach is attempted under the acquisition Doppler (AD) geometry, rather than a standard cylindrical reference system. It essentially utilizes one-dimensional azimuth Fourier domain processing, followed by range time-domain integration. As a result, the proposed algorithm is more computationally efficient compared to the time-domain method and yet maintains high accuracy. The effectiveness and efficiency of the proposed algorithm are demonstrated by numerical simulations.

This study examines the utility of cocollected, dual-wavelength, full-waveform lidar data to characterize vegetation and landscapes through the extraction of waveform features, such as total waveform energy, canopy energy distribution, and foliage penetration metrics. Assessments are performed using data collected in May 2014 over Monterey, California, using the Chiroptera dual-laser lidar mapping system from Airborne Hydrography AB. Both full-waveform and discrete return data were collected simultaneously at green (532 nm) and near-infrared (NIR) (1064 nm) wavelengths; however, the two channels are operated independently at different pulse repetition frequencies, thus measurements are not spatially coincident. A voxelization approach is employed to generate pseudowaveforms for each wavelength along vertical columns in a regularly spaced grid, such that spectral waveform properties can be evaluated independently of spatial variations resulting from instrumentation configuration and collection scenario. The pseudowaveforms are parameterized and extracted parameters are mapped to raster layers, which are then used as inputs to a random forest classifier to predict land cover classifications across the survey area. In comparison to independent classification results for the two wavelength channels, the combination of the NIR and green response provided an improvement in overall classification accuracy of up to 6%. This effort presents the methodology associated with the voxelization approach and the exploitation of the pseudowaveform features, while illustrating a potential utility for geospatial classification using multiple wavelengths.

Tensor greedy algorithms are widely used in the compressive sensing of multidimensional data such as hyperspectral (HS) images. However, due to their single-atom selection approach, the traditional tensor greedy algorithms sometimes cannot provide a satisfactory speed. We propose an algorithm, containing two modes, to select multiatoms instead of selecting a single atom during each iteration. The first one is an intuitive mode which means to select a fixed number of atoms in each iteration, in order to speed up the reconstruction. As an improvement, we develop the second mode in which the number of selected atoms for each iteration decreases gradually. A larger number of atoms is selected in the former iterations aiming to obtain enough speed while a smaller number in later iterations aims to guarantee the reconstruction accuracy. We have also studied the effect of parameters in the algorithm on reconstruction and determined the proper parameter ranges. The performance of the proposed algorithm is verified by both synthetic data and HS image simulations. Moreover, we compare the proposed algorithm with a traditional tensor greedy algorithm, demonstrating that the reconstruction speed is greatly improved without compromising accuracy.

The presented work proposes fusion of panchromatic and multispectral images in a shearlet domain. The proposed fusion rules rely on the regional considerations which makes the system efficient in terms of spatial enhancement. The luminance hue saturation-based color conversion system is utilized to avoid spectral distortions. The proposed fusion method is tested on Worldview2 and Ikonos datasets, and the proposed method is compared against other methodologies. The proposed fusion method performs well against the other compared methods in terms of subjective and objective evaluations.

A performance metric for infrared and visible image fusion is proposed based on Weber’s law. To indicate the stimulus of source images, two Weber components are provided. One is differential excitation to reflect the spectral signal of visible and infrared images, and the other is orientation to capture the scene structure feature. By comparing the corresponding Weber component in infrared and visible images, the source pixels can be marked with different dominant properties in intensity or structure. If the pixels have the same dominant property label, the pixels are grouped to calculate the mutual information (MI) on the corresponding Weber components between dominant source and fused images. Then, the final fusion metric is obtained via weighting the group-wise MI values according to the number of pixels in different groups. Experimental results demonstrate that the proposed metric performs well on popular image fusion cases and outperforms other image fusion metrics.

The availability of polarimetric synthetic aperture radar (PolSAR) images has increased, and consequently, the classification of such images has received immense attention. Among different classification methods in the literature, it is possible to distinguish them according to learning paradigm and approach. Unsupervised methods have as advantage the independence of labeled data for training. Regarding the approach, image classification can be performed based on its individual pixels or on previously identified regions in the image. Previous studies verified that the region-based classification of PolSAR images using stochastic distances can produce better results in comparison with the pixel-based. Faced with the independence of training data by unsupervised methods and the potential of the region-based approach with stochastic distances, this study proposes a version of the unsupervised K-means algorithm for PolSAR region-based classification based on stochastic distances. The Bhattacharyya stochastic distance between Wishart distributions was adopted to measure the dissimilarity among regions of the PolSAR image. Additionally, a measure was proposed to compare unsupervised classification results. Two case studies that consider real and simulated images were conducted, and the results showed that the proposed version of K-means achieves higher accuracy values in comparison with the classic version.

Atmospheric correction is a necessary step for deriving surface geophysical parameters. The aim of this paper is to study the atmospheric correction of Landsat-8 imageries released by the Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences. A look-up table (LUT)-based atmospheric correction method on Landsat-8 OLI is proposed. The LUT is generated with 6S model with inputs including total atmospheric water vapor content, ozone, and aerosol optical thickness (AOT) from MODIS atmospheric level 2 products. The conventional method to build up the atmospheric parameter LUT usually only takes part of the factors (e.g., AOT) into consideration, whereas it is not applicable in the atmospheric correction using per pixel of MODIS products as input atmospheric parameters. Thus, a five-dimensional LUT, which considers most input parameters, is built up and has high universality for the Landsat-8 OLI sensor. Finally, spectral analysis, comparison to U.S. Geological Survey-released surface reflectance (SR) products, and field observations are used to validate the quality of the model-computed SR. The validation results indicate that the proposed method can effectively generate accurate and reliable SR results, although there is an overcorrected problem in the costal blue region when the AOT value is very high.

Image matching has been a central issue in the field of computer vision and image processing for decades. It is normally based on the maximum similarity between two given images. We propose a similarity measure criterion based on structural similarity (SSIM) for image matching. We use the similarity measure criterion to compute the similarity of the feature points of two images to obtain the matching points. The results show that the correct matching rate of the proposed similarity measure criterion is higher than that of the normalized correlation coefficient. As for the mismatches in the initial set of matches, we use the proposed algorithm to compute the mean structural similarity (MSSIM) index of their neighborhood window image and eliminate those whose values of the MSSIM index are below the threshold. Then we use the geometric distribution of corresponding points in the image space to eliminate the mismatches that are difficult to eliminate using the SSIM algorithm. The experimental result shows that the proposed algorithm is superior to the random sample consensus algorithm in terms of mismatch detection and computational time.

Narrowband radar imaging algorithms based on the micro-Doppler (m-D) characteristics of precessional targets will lose efficacy since specular scattering makes the spectrogram discontinuous and unrecognizable. Specular scattering occurs under the condition that the incident angle of wave is perpendicular to the target surface. To image the precessional target under specular scattering, we propose a narrowband imaging algorithm via complex-valued empirical-mode decomposition-time frequency distribution-general Radon transform (CEMD-TFD-GRT) after analyzing the m-D characteristics of specular scattering centers. The CEMD first decomposes the echo into several components according to their spectrums varying from high to low so that the Doppler spectrums suppressed by the specular scattering can be recognized. Then, TFD-GRT is used to extract the parameters of m-D curves from which the positions of scattering centers can be reconstructed. In addition, the computation complexity of CEMD-TFD-GRT is analyzed and the Cramer–Rao low bounds for the coordinates estimation are derived. The experiment results with anechoic chamber data demonstrate that the scattering centers of precessional targets can be imaged with the proposed algorithm even when specular scattering occurs. The noise influence on the proposed algorithm is also presented with the experiments.

Stepped frequency continuous wave ground penetrating radar (SFCW-GPR) systems are becoming increasingly popular in the GPR community due to the wider dynamic range and higher immunity to radio interference. The traditional back-projection (BP) algorithm is preferable for SFCW-GPR imaging in layered mediums scenarios for its convenience and robustness. However, the existing BP imaging algorithms are usually very computationally intensive, which limits their practical applications to SFCW-GPR imaging. To solve the above problem, a fast SFCW-GPR BP imaging algorithm based on nonuniform fast Fourier transform (NUFFT) technique is proposed in this paper. By reformulating the traditional BP imaging algorithm into the evaluations of NUFFT, the computational efficiency of NUFFT is exploited to reduce the computational complexity of the imaging reconstruction. Both simulation and experimental results have verified the effectiveness and improvement of computational efficiency of the proposed imaging method.

It is a broadly established belief that the segmentation result significantly affects subsequent image classification accuracy. However, the actual correlation between the two has never been evaluated. Such an evaluation would be of considerable importance for any attempts to automate the object-based classification process, as it would reduce the amount of user intervention required to fine-tune the segmentation parameters. We conducted an assessment of segmentation and classification by analyzing 100 different segmentation parameter combinations, 3 classifiers, 5 land cover classes, 20 segmentation evaluation metrics, and 7 classification accuracy measures. The reliability definition of segmentation evaluation metrics as indicators of land cover classification accuracy was based on the linear correlation between the two. All unsupervised metrics that are not based on number of segments have a very strong correlation with all classification measures and are therefore reliable as indicators of land cover classification accuracy. On the other hand, correlation at supervised metrics is dependent on so many factors that it cannot be trusted as a reliable classification quality indicator. Algorithms for land cover classification studied in this paper are widely used; therefore, presented results are applicable to a wider area.

As random stepped-frequency chirp (RSFC) signal is used in wide-band radar applications such as synthetic aperture radar (SAR) and inverse SAR. RSFC has advantages over the linear stepped-frequency chirp, including suppressing the range ambiguity, decoupling the range-Doppler coupling, and reducing the signal interference. RSFC is usually descrambled and then fed to the inverse fast Fourier transform (IFFT) to achieve a coherent integration as well as a high-resolution range. However, this method needs frequency descrambling and accurate velocity pre-estimation for moving target detection. We propose a coherent integration method based on time-dechirping for bistatic radar. This method can detect moving targets without frequency descrambling or accurate velocity pre-estimation. This paper first models the target echo mathematically and outlines the difficulties associated with the processing of IFFT for RSFC. Then the detailed principles of the proposed method are introduced and the flowchart is given. Finally, numerical simulations are conducted to verify the effectiveness of the proposed method and show its detecting ability in the presence of noise.

In order to leverage computational complexity and avoid information losses, “big data” analysis requires a new class of algorithms and methods to be designed and implemented. In this sense, information theory-based techniques can play a key role to effectively unveil change and anomaly patterns within big data sets. A framework that aims at detecting the anomaly patterns of a given dataset is introduced. The proposed method, namely PROMODE, relies on a representation of the given dataset performed by means of undirected bipartite graphs. Then the anomalies are searched and detected by progressively spanning the graph. The proposed architecture delivers a computational load that is less than that carried by typical frameworks in literature, so that PROMODE can be considered as a valid algorithm for efficient detection of change patterns in remotely sensed big data.

This paper presents a practical method for landslide detection, using single-temporal Landsat8 image at a spatial resolution of 30 m, which is publicly available. The method introduces the concept “saliency” to express potential landslide regions in the image. Based on the saliency calculations, landslides are further extracted by object-based contours using morphological operations. The experimental area covers 2 deg×2 deg, which is more practical for most cases, and the performance validates the efficiency and robustness of this method in practical applications. The overall accuracy in terms of landslide/background classification reaches 99.87%, indicating the proposed method is viable. Even though the commission error and omission error of the detected landslides are both above 50%, the proposed method is able to remove ∼99.9% background, thus reducing manual interpretation to a large extent. Moreover, since it does not need training data or many experienced parameters, it is much easier to be applied to other cases. This paper moves forward a step in landslide detection toward practical applications, such as emergency response.

With the emergence of very high-resolution airborne synthetic aperture radar systems, it is necessary to reinvestigate these proposed methods with respect to their despeckling performances. As for the very high resolution polarimetric synthetic aperture radar (PolSAR) data, the presumption that the resolution cell is much larger than the radar wavelength becomes ineffective. Therefore, some classic and new filters are thoroughly reviewed. For the evaluation of speckle filters, both indicators for polarimetric information and spatial information are listed. The absolute relative bias is introduced, with the purpose of measuring the filtering performance concerning the indicators for polarimetric information. Moreover, the ratio of half power point width is employed to quantitatively assess the degree of point target preservation. A series of experiments are carried out based on the real PolSAR imagery which is obtained from an uninhabited aerial vehicle synthetic aperture radar system. It can be concluded that existing filters can only attain good performance with reference to part of the indicators. As regards very high-resolution PolSAR imagery, it is necessary to conceive more apposite new filters or make improved versions of the existing filters.

A time series algorithm was being presented that classifies vegetation in the Njoro Watershed, Kenya, according to the shapes of temporal normalized difference vegetation index (NDVI) profiles representing growing cycles for different vegetation. We present a two-step approach that includes noise reduction using discrete Fourier filtering and a clustering algorithm that uses the Fourier components of magnitude and phase to identify phenological differences. The classification considers possible variations in shape that may be imposed by climate, soil, topography, or human impacts. The primary input to the classification is a user-defined set of reference cycles to which pixels are assigned depending on a set of shape criteria. The output is a consistent classification of NDVI cycles representing vegetation classes with similar phenologies. The algorithm allows the creation of classification mosaics without typical boundary offsets and temporally comparable classification products. It identifies vegetation more accurately than single image classification methods, because it exploits the temporal variability in spectral reflectance due to phenological responses. We produced a classified land cover map at two hierarchical levels with five classes in a level I classification and seven classes in a level II classification that represents a refinement of the level I data. These map products compare favorably to previously published land cover maps that were developed using more standard supervised classification. The level I map has an overall accuracy of 94% compared to field data, while the level II map as an overall accuracy of 77%.

Bistatic inverse synthetic aperture radar (ISAR) is an efficient geometric configuration of ISAR and can provide complementary information to monostatic ISAR. However, bistatic ISAR suffers from a serious defocusing problem and distortion problem for maneuvering targets. We propose a method to solve these problems. First, cubic chirplet decomposition based on Keystone transform (KTCCD) is proposed to eliminate the defocusing problem. The echo is assumed as a combination of cubic chirplet atoms, and Keystone transformation is utilized to estimate the parameters for each atom. Then, a method based on distortion term estimation is proposed to eliminate the distortion problem. The distortion terms in certain range bins are expressed by the initial frequency and the chirp rate, and the distortion terms in other range bins are fitted by the ordinary least square algorithm. Simulation results validate the theoretical analysis and show the effectiveness of our method.

The alteration of surrounding rock is an important prospecting indicator in mineral exploration, but some important minerals are unclassified or misclassified when using hyperspectral remote sensing mineral recognition. A method for mineral recognition mapping was proposed. In this method, a decision tree discrimination rule was established based on the classification and regression tree data-mining algorithm and the absorption characteristics of field-measured spectra. Compared with spectral angle mapping and mixture-tuned matched filtering (MTMF), this method is shown to be efficient for mineral recognition mapping using hyperspectral images; its accuracy is 85.06%, which is greater than that of the MTMF method (83.91%). The advantages of the proposed method comprise the reduction of errors caused by the setting of the artificial threshold for mineral mapping and the lesser degree of difficulty in its training. Furthermore, the hierarchy structure of the decision tree in this method reflects the recognition process clearly, and the rule nodes are closely related to the spectra of the minerals; therefore, the advantage of this method is the interpretability of the results and the process. This method could be used for mineral recognition and classification using hyperspectral images.

Endmember extraction is a key step in hyperspectral unmixing. A new endmember extraction framework is proposed for hyperspectral endmember extraction. The proposed approach is based on the swarm intelligence (SI) algorithm, where discretization is used to solve the SI algorithm because pixels in a hyperspectral image are naturally defined within a discrete space. Moreover, a “distance” factor is introduced into the objective function to limit the endmember numbers which is generally limited in real scenarios, while traditional SI algorithms likely produce superabundant spectral signatures, which generally belong to the same classes. Three endmember extraction methods are proposed based on the artificial bee colony, ant colony optimization, and particle swarm optimization algorithms. Experiments with both simulated and real hyperspectral images indicate that the proposed framework can improve the accuracy of endmember extraction.

The image fusion process basically produces a high-resolution image by combining the superior features of a low-resolution spatial image and a high-resolution panchromatic image. Despite its common usage due to its fast computing capability and high sharpening ability, the intensity–hue–saturation (IHS) fusion method may cause some color distortions, especially when a large number of gray value differences exist among the images to be combined. This paper proposes a spatially adaptive IHS (SA-IHS) technique to avoid these distortions by automatically adjusting the exact spatial information to be injected into the multispectral image during the fusion process. The SA-IHS method essentially suppresses the effects of those pixels that cause the spectral distortions by assigning weaker weights to them and avoiding a large number of redundancies on the fused image. The experimental database consists of IKONOS images, and the experimental results both visually and statistically prove the enhancement of the proposed algorithm when compared with the several other IHS-like methods such as IHS, generalized IHS, fast IHS, and generalized adaptive IHS.

Classifying light detection and ranging (LiDAR) data into water and land points is an issue for the application of low-attitude airborne LiDAR (e.g., digital terrain model generation and river shoreline extraction). To solve the problem of distinguishing the water points from the land points in complex landscapes, an adaptive classification algorithm of water LiDAR point clouds is proposed, which consists of the following steps. First, the descriptors of local terrain slope and point density are designed by analyzing the characteristics of low-altitude airborne LiDAR water point clouds. Then Bayes’ theorem is introduced to establish membership functions of the elevation, slope, and density. Next, the adaptive weights of the individual membership functions are determined according to the t-test of the independent samples of water and land points. Finally, a classification model based on multifeature statistics is obtained, and the adaptive classification threshold of the model is determined by the probability density of the training samples. Typical experiments conducted in the middle-lower Yangtze River riparian zone indicate that water classification accuracies higher than 99% are obtained by this algorithm, even in complex landscapes with mudflats and inland plains.

Ionospheric scintillation can cause BeiDou system (BDS) signal amplitude fading and phase variation as the radio frequency (RF) signals pass through the ionosphere. With the accidental, sudden, and regional characteristics of ionospheric scintillation, it is difficult to effectively mitigate the impacts of ionospheric scintillation from a system perspective. To overcome this problem, this work addresses the mitigation of scintillation for the BDS signal. We propose an adaptive wavelet denoising unscented Kalman filter (WDUKF)-based carrier tracking algorithm to mitigate the adverse impacts of ionospheric scintillation. First, as WD can better characterize the nonstationary of signal, the non-Gaussian noise caused by scintillation can be filtered by designing a sliding window-based online WD filter. Second, taking the denoised integrations as an input, we employ the phase lock indicator (PLI)-based adaptive UKF carrier phase estimator to reduce the linearization approximate error of conventional KF-based algorithms. Through adjusting the measurement vector of UKF with different scintillation scenarios adaptively, the divergence problem of UKF can be mitigated. Simulation and real data experimental results demonstrate the validity of the proposed method, especially in the case of 36 dB-Hz in moderate scintillation, the phase jitter and probability of loss-of-lock decrease about 6 deg (40%) and 71%, respectively.

The Robinia pseudoacacia forest in the Yellow River delta of China has been planted since the 1970s, and a large area of dieback of the forest has occurred since the 1990s. To assess the condition of the R. pseudoacacia forest in three forest areas (i.e., Gudao, Machang, and Abandoned Yellow River) in the delta, we combined an estimation of scale parameters tool and geometry/topology assessment criteria to determine the optimal scale parameters, selected optimal predictive variables determined by stepwise discriminant analysis, and compared object-based image analysis (OBIA) and pixel-based approaches using IKONOS data. The experimental results showed that the optimal segmentation scale is 5 for both the Gudao and Machang forest areas, and 12 for the Abandoned Yellow River forest area. The results produced by the OBIA method were much better than those created by the pixel-based method. The overall accuracy of the OBIA method was 93.7% (versus 85.4% by the pixel-based) for Gudao, 89.0% (versus 72.7%) for Abandoned Yellow River, and 91.7% (versus 84.4%) for Machang. Our analysis results demonstrated that the OBIA method was an effective tool for rapidly mapping and assessing the health levels of forest.

This article proposes a symmetric sparse representation (SSR) method to extract pure endmembers from hyperspectral imagery (HSI). The SSR combines the features of the linear unmixing model and the sparse subspace clustering model of endmembers, and it assumes that the desired endmembers and all the HSI pixel points can be sparsely represented by each other. It formulates the endmember extraction problem into a famous program of archetypal analysis, and accordingly, extracting pure endmembers can be transformed as finding the archetypes in the minimal convex hull containing all the HSI pixel points. The vector quantization scheme is adopted to help in carefully choosing the initial pure endmembers, and the archetypal analysis program is solved using the simple projected gradient algorithm. Seven state-of-the-art methods are implemented to make comparisons with the SSR on both synthetic and real hyperspectral images. Experimental results show that the SSR outperforms all the seven methods in spectral angle distance and root-mean-square error, and it can be a good alternative choice for extracting pure endmembers from HSI data.

Point cloud registration is very important in three-dimensional (3-D) point cloud data processing as its results directly affect 3-D object reconstruction and other applications. Currently, there are many methods for point cloud registration, but these methods are not able to simultaneously solve the problem of both efficiency and precision. We propose a method of point cloud registration based on fast point feature histogram (FPFH), in which feature points are first extracted from the point cloud dataset according to FPFH and four point-to-point correspondences are found within some given constraints regarding their features, distances, and location relationships. Then, additional point pairs are added on the basis of the initial four point pairs until the number of point pairs satisfies the requirements for point cloud registration. Finally, a rigid transformation matrix is calculated from the correspondence of the point pairs. The results show that there is both a high efficiency and precision in most types of datasets when using this method for point cloud registration.

Sustainability of desert ecosystem is highly dependent upon water availability from different sources. Paleochannels are important sources of groundwater, and exploiting such resources involves their identification/mapping and subsequent investigation for fresh groundwater. A study in which multisensor (optical/infrared Landsat 8 OLI and active microwave Envisat ASAR) images of the Cholistan desert of Pakistan were processed and analyzed to identify and map Hakra River paleochannels is presented. Radiometrically corrected optical and synthetic aperture radar datasets were fused using principal components image fusion method. The paleochannels were extracted from the analysis of this fused output, and normalized difference vegetative index analysis of Landsat 8 OLI atmospheric corrected images was used as supporting information. Identification and alignment of an identified paleochannel was validated with geophysical ground measurements (electrical resistivity and conductivity surveys) and historical records. The presence of high apparent electrical resistivity with corresponding low soil water conductivity values intersects well with the paleochannels identified from the remote sensing data. The results were also confirmed with historical evidence such as old wells beside forts and proposed ground water harvesting sites. The proposed methodology in this study could be adopted in other parts of the world for mapping of paleochannels.

The high plateaus of Algeria is a critical region to policymakers in terms of social, economic, and infrastructure development. The main goal of the present work was to monitor the climatic drought and its impact on vegetation health across the Algerian high plateaus using remote sensing techniques. Vegetation health index (VHI) showed a clear drought in the western region of the study area. The results show practically three periods of drought were evident: October to December 2006, November to December 2009, and December 2012. Agreeable correlations among the obtained results using standard precipitation index for 3 months (SPI-3) and other satellite indicators such as temperature vegetation dryness index (TVDI) and VHI were obtained. TVDI and VHI agreed well with the ground-based observations from SPI-3; thus, these may serve as key and easily accessible indicators of drought. The research shows motivating results that decision makers can use to take timely corrective measures to minimize the reduction in agricultural production in drought prone areas.

Forest above-ground biomass (AGB) is an important indicator for understanding the global carbon cycle. It is hard to obtain a geographically and statistically representative AGB dataset, which is limited by unpredictable environmental conditions and high economical cost. A spatially explicit AGB reference map was produced by airborne LiDAR data and calibrated by field measurements. Three different sampling strategies were designed to sample the reference AGB, PALSAR backscatter, and texture variables. Two parametric and four nonparametric models were established and validated based on the sampled dataset. Results showed that random stratified sampling that used LiDAR-evaluated forest age as stratification knowledge performed the best in the AGB sampling. The addition of backscatter texture variables improved the parametric model performance by an R² increase of 21% and a root-mean-square error (RMSE) decrease of 10  Mg ha⁻¹. One of the four nonparametric models, namely, the random forest regression model, obtained comparable performance (R²=0.78, RMSE=14.95  Mg ha⁻¹) to the parametric model. Higher estimation errors occurred in the forest stands with lower canopy cover or higher AGB levels. In conclusion, incorporating airborne LiDAR and PALSAR data was proven to be efficient in upscaling the AGB estimation to regional scale, which provides some guidance for future forest management over cold and arid areas.

For the past several decades, scientists have been exploring the possibilities of decoding the complex physical mechanism of the earthquake process using satellites and ground-based technologies. In recent years, use of satellite-based technologies has been preferred, since monitoring a broader region and round-the-clock surveillance is possible using geostationary satellites. Using sensors mounted on satellites, it is possible to observe anomalous thermal variations in both the atmosphere and the ground. Measurement of anomalous variations in outgoing longwave radiation (OLR) is one such technology used to identify earthquake preparation zones. Anomalous variations in OLR have been observed in the vicinity of the epicenter, 3 to 30 days prior to impending earthquakes. We have analyzed the OLR scenario for earthquakes with magnitude greater than 7.0 that occurred during April 2016. Analysis has been done to find the relationship between OLR flux variations and the magnitude of the earthquakes. From the analysis, we have found that the recurrence frequency of OLR anomaly and cumulative energy flux of OLR can give us vital clues about the probable magnitude of impending earthquakes.

Cirrus clouds play an important role in the Earth’s radiation budget due to their high frequency of occurrence, nonspherical ice crystal formations, and variability in scattering/absorption characteristics. Mostly, tropical cirrus clouds are considered greenhouse modulators. Thus, the parameterization of tropical cirrus clouds in terms of their microphysical properties and the corresponding radiative effects are highly important for climate studies. For characterizing the radiative properties of cirrus clouds, which depend on the size, shape, and number of ice crystals, knowledge of the extinction coefficient (σ) and optical depth (τ) is necessary. σ provides information needed for understanding the influence of the scatterers on the radiative budget, whereas τ gives an indication of the composition and thickness of the cloud. Extensive research on tropical cirrus clouds has been carried out by using ground-based lidar (GBL) and satellite-based lidar systems. The characteristics of tropical cirrus clouds derived by using the data from the GBL system over the tropical site Gadanki (13.5° N, 79.2° E), India, during 2010 are presented. Some of the results are compared with those obtained by us from satellite-based cloud–aerosol lidar with orthogonal polarization observations of the cloud–aerosol lidar and infrared pathfinder satellite observation mission. It is observed that there is a strong dependence on some of the physical properties, such as occurrence height, cloud temperature, and geometrical thickness, and on the microphysical parameters in terms of extinction coefficient and optical depth. The correlation of both σ and τ with temperature is also observed.

Feature extraction and matching are two important steps in synthetic aperture radar automatic target recognition. This paper uses the binary target region as the feature and proposes a matching scheme for the target regions using binary morphological operations. The residuals between the testing target region and its corresponding template target regions are processed by the morphological opening operation. Then, a similarity measure is defined based on the residual remains to evaluate the similarities between different targets. Afterward, a Bayesian decision fusion is employed to fuse the similarities gained by different structuring elements to further enhance the recognition performance. The nonlinearity of the opening operation as well as the Bayesian decision fusion makes the proposed method robust to the nonlinear deformations of the target region. Experimental results on the moving and stationary target acquisition and recognition dataset demonstrate the validity of the proposed method.

We use measurements from the Cloud Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) to identify optically thin midlevel ice clouds and determine their occurrence and properties. We identify these clouds as having cloud top heights between 2 km above the ground and the tropopause plus 1 km, cloud base temperatures less than −10°C, and cloud top temperatures greater than −38°C. Globally, we find that these clouds occur at least 5% of the time and represent 7% of all tropospheric clouds detected by CALIPSO at a horizontal scale of 10 km. The cloud type fraction decreases with increasing horizontal scale, representing 4% of clouds at 25 km, 2% at 50 km, and less than 0.1% at 100 km. These clouds occur most commonly in the Arctic, most often in winter and least often in summer. During the winter, these clouds occur up to 15% of the time in the Arctic. We found five large (∼500  km) distinct clouds over the Arctic and investigated their meteorological conditions and radiative effects. We find that these thin midlevel ice clouds have a warming effect on the surface from 23 to 48  W/m². Our study highlights the importance of active satellite-based remote sensing in globally detecting and characterizing optically thin clouds.

Aquarius is a mission that aims to measure the sea surface salinity (SSS) from space to provide the global salinity for climate studies. Accurate estimation of SSS is useful for the hydrological cycle, oceanographic processes, and climate. Recently, the new version (V4) of Aquarius data releases with the improving of the quality of the data and achieving the mission accuracy requirement globally on monthly scale. The results of the comparison of recently released Aquarius V4 with its preceding version’s (version 3; V3) data on a monthly time scale from 2012 to 2014 periods are highlighted. Spatial distribution of mean SSS shows that both products capture the SSS variability very well. Aquarius V4 SSS showed a minor improvement over the Aquarius V3 SSS with less root-mean-square error over the central and eastern equatorial Indian Ocean and part of Bay of Bengal. This is the region where both versions have a large difference in mean SSS. The frequency distribution is also improved in Aquarius V4 compared to Aquarius V3 average over the different regions. However, both the versions overestimate/underestimate the frequency of low/high salinity values. We also used the ocean reanalysis data to show the improvement in Aquarius V4 using the triple-point analysis method.

The high dimensionality of hyperspectral images (HSIs) with significant levels of redundancy and noise is a pervasive problem that creates a heavy burden and many inconveniences in terms of accuracy and efficiency for all the subsequent applications. To address this problem, we proposed an adaptive-boundary adjustment-based groupwise band-categorization (BC) framework that classifies bands by the information contained within each band and redundant/noisy components without compromising the original content from the raw HSI. Since HSI labels are difficult to collect, an unsupervised spectral-spatial adaptive band-noise factor-based formulation is devised for BC based on the measurement of boundary movement/adjustment and band correlation. Robust clustering and boundary-adjustment-based methods involve two key issues: band correlation and band preference. Our proposed method contemplates band correlation and band synergy, and band preference is defined by the discriminative information contained in each band. According to different information-distribution levels, an adaptive mechanism is proposed to retain the spectral correlations in spectral dimensions and match the actual structure in spatial dimensions. The proposed method is based on the observation that clustering and boundary adjustment techniques are sensitive to noise and redundancy, which can effectively contribute to estimates and help classify the bands. Comprehensive experimental results on challenging airborne visible/infrared-imaging spectrometer datasets from Indian Pines, Moffett Field, and Salinas demonstrate the significant advancement in these two applications and show the reliability and effectiveness of the proposed framework without compromising the informative bands. Because of its usefulness, BC can be successfully applied to many practical applications of hyperspectral remote sensing. Effective application of the proposed framework is demonstrated by applying it on HSI classification and also for estimating the noise. The ability of the method to characterize the bands along with the noise estimation of HSIs is of significant benefit for subsequent remote-sensing techniques.

The idea of using commercially available cascade lasers for stand-off detection of alcohol vapors in moving cars is presented. Special attention is paid to the optical characteristics of the car windowpanes for the monitoring as well as for the reference laser beams. A special experimental setup was built to investigate the idea. It is shown that using interband cascade lasers operating at 3.45- and 3.59-μm wavelengths, the alcohol vapors inside a car can be successfully detected, even in cars with different windowpanes.

The extractive and construction industries rely heavily on accurate geospatial data to control position, location, alignment, and orientation of planned excavations. Recent advancements in the survey industry, through the use of terrestrial laser scanning, can now provide engineering teams with three-dimensional (3-D) data in unprecedented detail via georeferenced point clouds. Furthermore, equipment is now available that provides fully mobile automated mapping solutions, independent of satellite positioning, utilizing simultaneous localization and mapping. This paper evaluates the surveying capability of three fully mobile automated mapping solutions against a benchmark laser scanning survey undertaken at the underground Camborne School of Mines Test Mine facility. The study highlights that handheld automated mapping solutions, in which closed-loops can be formed, have the potential to provide quicker data collection and processing time, as well as the required accuracy for underground surveying applications. However, the automated solution was unable to produce the necessary point cloud density to identify low-angled discontinuities that may have a major safety implication, leading to potential rockfall.

We present a radar sensor that was designed to detect and image moving objects/targets on the other side of a wall. The radar sensor was composed of a linear array of Vivaldi antenna elements, an radio frequency (RF) switch, a microcontroller unit, and an RF transceiver. For the linear array, a total of eight antenna elements were used as sensors in synthetic aperture radar (SAR) configuration in the cross-range axis to improve the resolution in this dimension. Design steps of Vivaldi antenna elements and the entire linear array were presented. After the design, the prototyping procedure and the details of the radar sensor were given. Through-the-wall radar (TWR) imaging experiments were performed for stationary and moving targets using the assembled sensor. The resultant TWR images after these experiments were presented. During the image formation, a back-projection type image focusing algorithm was implemented and applied to increase the signal-to-noise ratio of the raw images. The constructed radar images demonstrated that our radar sensor could successfully detect and image both stationary and moving targets on the other side of the wall.

Ground-based synthetic aperture radar (GBSAR) is a powerful tool used in monitoring structures such as bridges and dams. However, despite the extremely short range of GBSAR interferometry, the atmospheric effects cannot be neglected. The permanent scatterer (PS) technique is an effective operational tool that utilizes a long series of SAR data and detects information with high accuracy. An algorithm based on the PS technique is developed in accordance with the phase model used in GBSAR interferometry. Atmospheric correction is carried out on a real campaign (Geheyan Dam, China). The atmospheric effects created using this method, which utilizes SAR data, can be effectively reduced compared to when plumb line data are used.

Considering the inevitable obstacles faced by the pixel-based clustering methods, such as salt-and-pepper noise, high computational complexity, and the lack of spatial information, a reweighted mass center based object-oriented sparse subspace clustering (RMC-OOSSC) algorithm for hyperspectral images (HSIs) is proposed. First, the mean-shift segmentation method is utilized to oversegment the HSI to obtain meaningful objects. Second, a distance reweighted mass center learning model is presented to extract the representative and discriminative features for each object. Third, assuming that all the objects are sampled from a union of subspaces, it is natural to apply the SSC algorithm to the HSI. Faced with the high correlation among the hyperspectral objects, a weighting scheme is adopted to ensure that the highly correlated objects are preferred in the procedure of sparse representation, to reduce the representation errors. Two widely used hyperspectral datasets were utilized to test the performance of the proposed RMC-OOSSC algorithm, obtaining high clustering accuracies (overall accuracy) of 71.98% and 89.57%, respectively. The experimental results show that the proposed method clearly improves the clustering performance with respect to the other state-of-the-art clustering methods, and it significantly reduces the computational time.

Classification of real-world remote sensing images is a challenging task because of complex spectral–spatial information with high-dimensional feature vectors. Most of the traditional classification approaches directly treat data as vectors, which usually results in a small sample size problem and abundant redundant information; thus, they inevitably degrade the performance of the classifier. To overcome the drawbacks, we take advantage of the benefits of local scatters and tensor representation and propose a framework for hyperspectral image (HSI) classification through combining local tensor discriminant analysis (LTDA) with spectral–spatial feature extraction. First, we use a well-known spectral–spatial feature extraction approach to extract abundant spectral–spatial features as feature tensors. Then, based on class label information, LTDA is used to eliminate redundant information and to extract discriminant feature tensors for the subsequent classification. Two real HSIs are used as experimental datasets. The obtained results indicate that the proposed method exhibits good performance, while using a small number of training samples.

An improved inversion algorithm for estimating the significant wave height (SWH) from marine X-band radar image sequences is developed. To retrieve the SWH from radar image sequences of heterogeneous nearshore wave field, the principal components of the radar image were extracted by empirical orthogonal function (EOF) analysis and rotated empirical orthogonal function (REOF) analysis, respectively. REOF method makes decomposition based on EOF method, which rotates in accordance with the principal of maximum variance orthogonal rotation. The basic idea of this method is that the relationship between the SWH and the standard deviation of principal components is linear. The change of SWH measured by the buoy ranges from 0.5 to 1.7 m. For validation, these two methods are, respectively, tested against the data from in situ buoys. The results are that the correlation coefficient between the wave parameters measured by the buoy and restored from radar image sequences using the EOF analysis is 0.85, and that using the REOF analysis is 0.90, whereas the root-mean-square error using EOF analysis is 0.20 m for SWH, and that using REOF analysis is 0.17 m. The results indicate that the effect of the REOF method is better than that of the EOF method.

This paper presents an unsupervised synthetic aperture radar (SAR) image change detection method based on improved bilateral filtering and fuzzy C means (FCM). Many previous approaches to change detection are based on a difference image. Unlike conventional approaches, based on difference images, our method demonstrates superior ability to reduce speckle noise and suppress background information, while still retaining edge information effectively. First, the two images are preprocessed using a Lee filter to remove some of the speckle noise. Second, we use the neighbor-log ratio and the Gauss-log ratio to produce initial change maps. Third, we use the improved bilateral to fuse the two change maps, to obtain an initial difference image. Next, we apply a median filter on the initial difference image, to obtain the final difference image. The above method makes full use of the field information, and it can effectively remove speckle noise while still preserving edge information. Finally, an improved FCM algorithm is used to cluster the denoised difference image. Denoising prior to clustering overcomes the main deficiency of conventional clustering algorithms, which is that they are noise sensitive. Empirical experiments, on three groups of SAR images, suggest that the proposed algorithm outperforms several other methods from the literature, in terms of noise suppression, accuracy, and lower change detection error rates.

A method for estimating the height of rice plants, using three-dimensional laser range data from point clouds, is proposed and assessed. Rice plant height (H) is estimated using a reference position at the top of the rice plant, avoiding the need to determine the ground position. Field experiments were performed with a SICK LMS 200 laser scanner in 2013 and 2014 on a test field with five different planting geometries. Percentile analysis identified the closest percentile to the top of the rice plant (p^t=1), with vertical distances at the first percentile unaffected by planting geometry. The plant bottom position was identified using three different percentile ranks (p^b=95, p^b  =80, and p^b  =70). Relative vertical distances (rD) were computed from the difference between the top and bottom positions of the rice plant. These correlated well with measured H, with slopes greater than 1.0. A greater number of stems in 2014 led to steeper slopes. Estimated H was more accurate when plant bottom positions were closer to the ground surface, and the best results were obtained with p^b=95 (r²>0.87; RMSE≈4  cm). Overall, H was typically 16.0 cm greater than   rD with p^b=95.

Change detection is of high practical value to hazard assessment, crop growth monitoring, and urban sprawl detection. A synthetic aperture radar (SAR) image is the ideal information source for performing change detection since it is independent of atmospheric and sunlight conditions. Existing SAR image change detection methods usually generate a difference image (DI) first and use clustering methods to classify the pixels of DI into changed class and unchanged class. Some useful information may get lost in the DI generation process. This paper proposed an SAR image change detection method based on neighborhood-based ratio (NR) and extreme learning machine (ELM). NR operator is utilized for obtaining some interested pixels that have high probability of being changed or unchanged. Then, image patches centered at these pixels are generated, and ELM is employed to train a model by using these patches. Finally, pixels in both original SAR images are classified by the pretrained ELM model. The preclassification result and the ELM classification result are combined to form the final change map. The experimental results obtained on three real SAR image datasets and one simulated dataset show that the proposed method is robust to speckle noise and is effective to detect change information among multitemporal SAR images.

The Tarim and Konqi Rivers in western China have experienced dramatic changes in streamflow and riparian vegetation due to climatic variability, land cover change, and water management including interbasin water transfers. To assess the extent and evolution of vegetation dynamics along these rivers, we use Landsat and MODIS images for land cover classification, spectral mixture analysis, and landscape phenology analysis. From 1998 to 2011, agriculture nearly tripled in extent, from 1376 to 3742  km². Natural riparian vegetation persisted in aggregate but experienced losses (to agriculture) in some areas while expanding into barren land elsewhere. Spectral mixture analysis suggests that interbasin water transfers from the Konqi to the Tarim River increased near-channel riparian vegetation on the Tarim at the expense of vegetation on the Konqi. A time-series of MODIS images reveals a pattern of increasing and decreasing greenness across the region, including loss of vegetation in distal regions that were formerly subject to sporadic seasonal flooding but now are cut off from their water supply due to water management. These results suggest that satellite remote sensing may play a valuable role in monitoring the effects of changing land use and hydrology on riparian systems in Central Asia and other arid regions.

Crop surface models (CSMs) representing plant height above ground level are a useful tool for monitoring in-field crop growth variability and enabling precision agriculture applications. A semiautomated system for generating CSMs was implemented. It combines an Android application running on a set of smart cameras for image acquisition and transmission and a set of Python scripts automating the structure-from-motion (SfM) software package Agisoft Photoscan and ArcGIS. Only ground-control-point (GCP) marking was performed manually. This system was set up on a barley field experiment with nine different barley cultivars in the growing period of 2014. Images were acquired three times a day for a period of two months. CSMs were successfully generated for 95 out of 98 acquisitions between May 2 and June 30. The best linear regressions of the CSM-derived plot-wise averaged plant-heights compared to manual plant height measurements taken at four dates resulted in a coefficient of determination R² of 0.87 and a root-mean-square error (RMSE) of 0.08 m, with Willmott’s refined index of model performance d_r equaling 0.78. In total, 103 mean plot heights were used in the regression based on the noon acquisition time. The presented system succeeded in semiautomatedly monitoring crop height on a plot scale to field scale.

High-dimensional data applications have earned great attention in recent years. We focus on remote sensing data analysis on high-dimensional space like hyperspectral data. From a methodological viewpoint, remote sensing data analysis is not a trivial task. Its complexity is caused by many factors, such as large spectral or spatial variability as well as the curse of dimensionality. The latter describes the problem of data sparseness. In this particular ill-posed problem, a reliable classification approach requires appropriate modeling of the classification process. The proposed approach is based on a hierarchical clustering algorithm in order to deal with remote sensing data in high-dimensional space. Indeed, one obvious method to perform dimensionality reduction is to use the independent component analysis process as a preprocessing step. The first particularity of our method is the special structure of its cluster tree. Most of the hierarchical algorithms associate leaves to individual clusters, and start from a large number of individual classes equal to the number of pixels; however, in our approach, leaves are associated with the most relevant sources which are represented according to mutually independent axes to specifically represent some land covers associated with a limited number of clusters. These sources contribute to the refinement of the clustering by providing complementary rather than redundant information. The second particularity of our approach is that at each level of the cluster tree, we combine both a high-level divisive clustering and a low-level agglomerative clustering. This approach reduces the computational cost since the high-level divisive clustering is controlled by a simple Boolean operator, and optimizes the clustering results since the low-level agglomerative clustering is guided by the most relevant independent sources. Then at each new step we obtain a new finer partition that will participate in the clustering process to enhance semantic capabilities and give good identification rates.

This paper develops an automatic method to identify pavement layer properties from ground penetrating radar (GPR) data collected at traffic speed. The GPR system is operated at a center frequency of 2 GHz with a penetration depth of 60 cm in common road materials. Features include the capability of collecting up to 1000  traces/s, a large dynamic range, and compacted packaging. Using a four-channel GPR system, a large amount of data are collected at traffic speed on urban roads for over 200 lane miles, providing a dense spatial coverage. The GPR data contain information about the pavement layer properties, including layer interface, dielectric constant, and layer thickness. Using cross correlation and Hilbert transform algorithms, the pavement layer properties are identified from the large GPR data sets automatically and efficiently. The method has been successfully demonstrated in engineering applications for the accurate estimation of the layer thickness with excellent repeatability. Moreover, thickness data from different radar channels at the same location are used for transversal profile prediction. By searching abnormal variations of layer properties and amplitude of reflection signals, features and/or possible distresses in surface and subsurface, such as full-depth asphalt patches and prepothole conditions, can be detected.

With the rapid development of remote sensing, multiple techniques are now capable of producing digital elevation models (DEMs), such as photogrammetry, Light Detection and Ranging (LiDAR), and interferometric synthetic aperture radar (InSAR). Satellite-derived InSAR DEMs are particularly attractive due to their advantages of large spatial extents, cost-effectiveness, and less dependence on the weather. However, several complex factors may limit the quality of derived DEMs, e.g., the inherited errors may be nonlinear and spatially variable over an entire InSAR pair scene. We propose a segmentation-based coregistration approach for generating accurate InSAR DEMs over large areas. Two matching algorithms, including least squares matching and iterative closest point, are integrated in this approach. Three Advanced Land Observing Satellite Phased Array type L-band Synthetic Aperture Radar (ALOS PALSAR) InSAR DEMs are evaluated, and their root mean square errors (RMSEs) improved from 17.87 to 9.98 m, 51.94 to 15.80 m, and 27.12 to 12.26 m. Compared to applying a single global matching strategy, the segmentation-based strategy further improved the RMSEs of the three DEMs by 3.27, 13.01, and 9.70 m, respectively. The results clearly demonstrate that the segmentation-based coregistration approach is capable of improving the geodetic quality of InSAR DEMs.

Fuel structural characteristics affect fire behavior including fire intensity, spread rate, flame structure, and duration, therefore, quantifying forest fuel structure has significance in understanding fire behavior as well as providing information for fire management activities (e.g., planned burns, suppression, fuel hazard assessment, and fuel treatment). This paper presents a method of forest fuel strata classification with an integration between terrestrial light detection and ranging (LiDAR) data and geographic information system for automatically assessing forest fuel structural characteristics (e.g., fuel horizontal continuity and vertical arrangement). The accuracy of fuel description derived from terrestrial LiDAR scanning (TLS) data was assessed by field measured surface fuel depth and fuel percentage covers at distinct vertical layers. The comparison of TLS-derived depth and percentage cover at surface fuel layer with the field measurements produced root mean square error values of 1.1 cm and 5.4%, respectively. TLS-derived percentage cover explained 92% of the variation in percentage cover at all fuel layers of the entire dataset. The outcome indicated TLS-derived fuel characteristics are strongly consistent with field measured values. TLS can be used to efficiently and consistently classify forest vertical layers to provide more precise information for forest fuel hazard assessment and surface fuel load estimation in order to assist forest fuels management and fire-related operational activities. It can also be beneficial for mapping forest habitat, wildlife conservation, and ecosystem management.

Robust reflectance-based skin detection is a potentially powerful tool for security and search and rescue applications, especially when applied to video. However, to be useful it must be able to account for the variations of human skin, as well as other items in the environment that could cause false detections. This effort focused on identifying a robust skin detection scheme that is appropriate for video application. Skin reflectance was modeled to identify unique skin features and compare them to potential false positive materials. Based on these comparisons, specific wavelength bands were selected and different combinations of two and three optical filters were used for actively identifying skin, as well as identifying and removing potential false positive materials. One wavelength combination (1072/1250  nm) was applied to video using both single- and dual-camera configurations based on its still image performance, as well as its appropriateness for video application. There are several important factors regarding the extension of still image skin detection to video, including light available for detection (solar irradiance and reflectance intensity), overall intensity differences between different optical filters, optical component light loss, frame rate, time lag when switching between filters, image coregistration, and camera auto gain behavior.

Accurate estimates are emerging with technological advances in remote sensing, and the triangle method has demonstrated to be a useful tool for the estimation of evaporative fraction (EF). The purpose of this study was to estimate the EF using the triangle method at the regional level. We used data from the Moderate Resolution Imaging Spectroradiometer orbital sensor, referring to indices of surface temperature and vegetation index for a 10-year period (2002/2003 to 2011/2012) of cropping seasons in the state of Paraná, Brazil. The triangle method has shown considerable results for the EF, and the validation of the estimates, as compared to observed data of climatological water balance, showed values >0.8 for modified “d” of Wilmott and R2 values between 0.6 and 0.7 for some counties. The errors were low for all years analyzed, and the test showed that the estimated data are very close to the observed data. Based on statistical validation, we can say that the triangle method is a consistent tool, is useful as it uses only images of remote sensing as variables, and can provide support for monitoring large-scale agroclimatic, specially for countries of great territorial dimensions, such as Brazil, which lacks a more dense network of meteorological ground stations, i.e., the country does not appear to cover a large field for data.

Chi-squared transform (CST), as a statistical method, can describe the difference degree between vectors. The CST-based methods operate directly on information stored in the difference image and are simple and effective methods for detecting changes in remotely sensed images that have been registered and aligned. However, the technique does not take spatial information into consideration, which leads to much noise in the result of change detection. An improved unsupervised change detection method is proposed based on spatial constraint CST (SCCST) in combination with a Markov random field (MRF) model. First, the mean and variance matrix of the difference image of bitemporal images are estimated by an iterative trimming method. In each iteration, spatial information is injected to reduce scattered changed points (also known as “salt and pepper” noise). To determine the key parameter confidence level in the SCCST method, a pseudotraining dataset is constructed to estimate the optimal value. Then, the result of SCCST, as an initial solution of change detection, is further improved by the MRF model. The experiments on simulated and real multitemporal and multispectral images indicate that the proposed method performs well in comprehensive indices compared with other methods.

Monitoring crop phenology is essential for evaluating crop productivity and crop management. Remote sensing provides an efficient way to monitor crop phenological metrics at a large-scale. However, the widely used AVHRR and MODIS images are less reliable at a small-scale and in areas with heterogeneous land covers, such as the patchy cropland in South Central China. Therefore, this study analyzed the spatial and temporal variations of winter wheat phenology in South Central China, using enhanced vegetation index (EVI) time series predicted by a spatio-temporal fusion method that combines information from Landsat and MODIS images. The 13-year predicted EVI showed a close correspondence with the EVI derived from original Landsat images. Start of season (SOS), peak greenness, and end of season (EOS) were derived from the predicted EVI time series. The comparison with ground observations showed that the differences between the predicted phenological metrics and observations were usually within seven days. The length of the growing season demonstrated high spatial heterogeneity over the study area and the spatial patterns varied from year to year. The phenological dates did not show obvious increasing or decreasing trends through 13 years. The length of the growing season in the study area was positively correlated with precipitation, but the duration from SOS to peak greenness and the duration from peak greenness to EOS were strongly and negatively correlated with hours of sunshine.

Information on land use and land cover changes is considered as a foremost requirement for monitoring environmental change. Developing change detection methodology in the remote sensing community is an active research topic. However, to the best of our knowledge, no research has been conducted so far on the application of random forest regression (RFR) and support vector regression (SVR) for natural hazard change detection from high-resolution optical remote sensing observations. Hence, the objective of this study is to examine the use of RFR and SVR to discriminate between changed and unchanged areas after a tsunami. For this study, RFR and SVR were applied to two different pilot coastlines in Indonesia and Japan. Two different remotely sensed data sets acquired by Quickbird and Ikonos sensors were used for efficient evaluation of the proposed methodology. The results demonstrated better performance of SVM compared to random forest (RF) with an overall accuracy higher by 3% to 4% and kappa coefficient by 0.05 to 0.07. Using McNemar’s test, statistically significant differences (Z≥1.96), at the 5% significance level, between the confusion matrices of the RF classifier and the support vector classifier were observed in both study areas. The high accuracy of change detection obtained in this study confirms that these methods have the potential to be used for detecting changes due to natural hazards.

The research is focused on the study of the applicability of remote sensing techniques (specifically using ASTER data) for marine environmental analysis, relative to the determination of the surface temperatures. Using a multitemporal approach, two images (classified as level-1B), acquired in August 2000 and in 2005, were considered. The thermal maps were realized by means of the emissivity spectral normalization method, defining the thermal gradients in the area under investigation. The spatial and temporal anomalies related to the temperature distribution were highlighted; these anomalies represent an important parameter for the identification of probable groundwater pollution and soil contamination. To define the map of the surface temperature using thermal infrared ASTER channels, a numerical model dedicated to ASTER data was implemented during the experimentation. This model is based on the principle of interpolation of the type “least square interpolation (linear),” and it implies a reduction in the number of unknowns to obtain an acceptable solution to the problem. This experimental model has provided good results in the phase of implementation and in the tests on a synthetic image that were simulated in the laboratory. However, further verifications and modifications are necessary for the processing of real ASTER images.

The estimation of atmospheric water vapor with high resolution is important for operational weather forecasting, climate monitoring, atmospheric research, and numerous other applications. The 40  m×40  m and 30  m×30  m differential precipitable water vapor (ΔPWV) maps are generated with C- and L-band synthetic aperture radar interferometry (InSAR) images over Shanghai, China, respectively. The ΔPWV maps are accessed via comparisons with the spatiotemporally synchronized PWV measurements from the European Centre for Medium-Range Weather Forecasts Interim reanalysis at the finest resolution and global positioning system observations, respectively. Results reveal that the ΔPWV maps can be estimated from both C- and L-band InSAR images with an accuracy of better than 2.0 mm, which, therefore, demonstrates the ability of InSAR observations at both C- and L-band to detect the water vapor distribution with high spatial resolution.

This PDF file contains the errata for “JRS Vol. 10 Issue 04 Paper JARS-2016-1117-ERR” for JRS Vol. 10 Issue 04