Hyperspectral imaging has been used in a variety of geological applications since its advent in the 1970s. In the last few decades, different techniques have been developed by geologists to analyze hyperspectral data in order to quantitatively extract geological information from the high-spectral-resolution remote sensing images. We attempt to review and update various steps of the techniques used in geological information extraction, such as lithological and mineralogical mapping, ore exploration, and environmental geology. The steps include atmospheric correction, dimensionality processing, endmember extraction, and image classification. It is identified that per-pixel and subpixel image classifiers can generate accurate alteration mineral maps. Producing geological maps of different surface materials including minerals and rocks is one of the most important geological applications. The hyperspectral images classification methods demonstrate the potential for being used as a main tool in the mining industry and environmental geology. To exemplify the potential, we also include a few case studies of different geological applications.

In hyperspectral remote sensing imagery, the sensor, atmosphere, topography, and other factors often bring about some degradations, such as noise, haze, clouding, and shadowing. Due to inevitable tradeoff between spatial resolution and spectral resolution, low spatial details of hyperspectral images (HSIs) also limit the range of potential applications. Compensating for these degradations through quality improvement is a key preprocessing step in the exploitation of HSIs. A comprehensive analysis of the quality improvement techniques for HSIs is presented. The closely connected techniques, such as denoising, destriping, dehazing, cloud removal, and super-resolution, are linked as a whole by a general reconstruction model in a variational framework. Furthermore, we classify the methods into four categories according to their processing strategies for HSIs, including single-channel prior-based model, cross-channel prior-based model, tensor-based model, and data-driven prior-based model. Then, for several specific tasks, we briefly introduce their architectures of quality improvement, which combine different models and available complementary information from other spectral bands and/or temporal/sensor images. Some experimental results in different tasks are presented to show the effect of variational framework and draw some meaningful conclusions. Finally, some advantages on variational framework are discussed, and several promising directions are provided to serve as guidelines for future work.

The guest editors introduce the Special Section on Satellite Remote Sensing for Disaster Monitoring and Risk Assessment, Management, and Mitigation.

This paper seeks to identify opportunities to integrate Earth observations (EO) into flood forecast-based early action and propose future directions for research and collaboration between EO and humanitarian communities. Forecast-based early action (FbA) is an approach to shift disaster response toward anticipation to mitigate impacts to at-risk communities; however, timely and accurate information is needed in the development of data-based triggers and thresholds for action. Therefore, this paper considers the readiness of a wide range of EO for flood monitoring and forecasting in the design, operations, and evaluation phases of FbA. The most significant opportunities for EO to inform FbA efforts lie in the design and evaluation phases, as EO can aid in the development of impact-based triggers. The EO products most readily applicable include precipitation, streamflow estimates, and exposure mapping, and those requiring the greatest amount of further research include vulnerability and impact assessments. This paper identifies collaboration opportunities for the EO and humanitarian communities to create tailored products, such as overlays combining flood extents with exposure maps. Such collaboration opportunities can be fostered by open data sharing, data verification efforts, and incentives for supporting boundary organizations capable of enabling the use of EO for FbA.

Remote sensing using synthetic aperture radar (SAR) is an important tool for emergency flood incident management. At present, operational services are mainly aimed at flood mapping in rural areas, as mapping in urban areas is hampered by the complicated backscattering mechanisms occurring there. A method for detecting flooding at high resolution in urban areas that may contain dense housing is presented. This largely uses remotely sensed data sets that are readily available on a global basis, including open-access Sentinel-1 SAR data, the WorldDEM digital surface model (DSM), and open-access World Settlement Footprint data to identify urban areas. The method is a change detection technique that locally estimates flood levels in urban areas. It searches for increased SAR backscatter in the post-flood image due to double scattering between water (rather than unflooded ground) and adjacent buildings, and reduced SAR backscatter in areas away from high slopes. Areas of urban flooding are detected by comparing an interpolated flood level surface to the DSM. The method was tested on two flood events that occurred in the UK during the storms of Winter 2019–2020. High urban flood detection accuracies were achieved for the event in moderate density housing. The accuracy was reduced for the event in dense housing, when street widths became comparable to the DSM resolution, though it would still be useful for incident management. The method has potential for operational use for detecting urban flooding in near real-time on a global basis.

Especially in recent years, studies to determine the effects of natural disasters from satellite images have been very popular. The destruction caused by the disaster and the early detection of the affected structures are of great importance for the establishment of the precautionary measures and the right action plan. However, studies in this area are mostly made observationally and as a result, desired results cannot be achieved. On the other hand, the introduction of machine learning-based detection methods is very promising. In this study, a damaged building detection method based on convolutional neural networks (CNN) is proposed. Unlike similar studies, the hyperparameters of the CNN are optimized using Bayesian optimization algorithm to obtain more accurate and reliable detection results. The testing and validation results performed with a large number of images reveal the robustness of the proposed method. In addition, the performance evaluation measures obtained from the balanced and unbalanced testing datasets solidified the success of the optimized CNN model.

Increasing demand for groundwater for agricultural and industrial needs puts pressure on the water table, especially in arid and semiarid climates. In the Kabudar Ahang plain, in Hamedan province, groundwater overexploitation for agricultural and industrial application has increased significantly over recent decades. The InSAR technique is used to monitor the subsidence induced by groundwater overexploitation in the Kabudar Ahang plain from 2003 to 2010. For this purpose, three different Envisat tracks are used. Interferogram stacking and time-series analysis are performed to study the short-term and long-term behavior of the subsidence. Interferogram time-series analysis from 2003 to 2006 estimates an average subsidence of 19 cm per year with a maximum of 25 cm. The maximum rate of 27 and 28 cm per year is estimated from interferogram stacking in 2004 to 2005 and 2007 to 2010, respectively. The results are further compared with the geological and hydrological information to investigate the relation between the subsidence and groundwater level variations. A high correlation is found in areas affected by subsidence and the rate of water level drop in 10 piezometric wells. The results show that a large part of the plain subsides, mostly in the south and southeast where the aquifer is thicker and in the areas with the highest water table dropdown of about 100 m. With the continued use of groundwater resources, subsidence is another hazard in this plain in addition to the sinkholes.

Drought is one of the most serious natural disasters affecting global food security and human life. Traditional drought monitoring is based on the impact of water deficit on crop morphology. Due to the time lag of traditional drought monitoring based on vegetation index and land surface temperature, crop yield has been adversely affected when drought detection is delayed. Early drought warning is an effective way to prevent disasters and reduce damage. An early warning model, using the percentage of anomaly vegetation and fluorescence (PAVF) was proposed to determine drought risk level and development trends. The principle of PAVF is to analyze the change in the percentage of anomaly vegetation (PAV) using the normalized difference vegetation index (NDVI), from the moderate-resolution imaging spectroradiometer sensor, and percentage anomaly fluorescence (PAF), from the Global Ozone Monitoring Experiment. A standard score of seasonal cumulative NDVI and Sun-induced chlorophyll fluorescence based on the data from 2007 to 2017 was used as a proxy for the evaluation index. Because PAV and PAF have different sensitivities to monitoring vegetation physiological changes, these parameters can reflect the occurrence and development trend of drought from different perspectives. Drought risk analysis in Xilingol League of China in 2018 demonstrated that the drought warning model could depict the evolution process of agricultural drought and can efficiently predict the occurrence of drought in the next half month. The proposed PAVF provides an available and innovative way to address the complexity of agricultural drought warnings.

Guest Editors Shen-En Qian, Robert O. Green, and Antonio J. Plaza introduce the Special Section on Hyperspectral Remote Sensing and Imaging Spectrometer Design.

Hyperspectral image (HSI) classification has many applications in different diverse research fields. We propose a method for HSI classification using principal component analysis (PCA), 2D spatial convolution, and support vector machine (SVM). Our method takes advantage of correlation in both spatial and spectral domains in an HSI data cube at the same time. We use PCA to reduce the dimensionality of an HSI data cube. We then perform spatial convolution to the dimension-reduced data cube once and then to the convolved data cube for the second time. As a result, we have generated two convolved PCA output data cubes in a multiresolution way. We feed the two convolved data cubes to SVM to classify each pixel to one of the known classes. Experiments on three widely used hyperspectral data cubes (i.e., Indian Pines, Pavia University, and Salinas) demonstrate that our method can improve the classification accuracy significantly when compared to a few existing methods. Our method is relatively fast in terms of central processing unit computational time as well.

Remote sensing image pan-sharpening is the fusion of multispectral (MS) image and panchromatic (PAN) image into fused image with MS and high-resolution information features. In recent years, deep learning technology has been widely used in abounding fields such as text and image recognition. Inspired by this, we proposed a multistream deep ResCNN (RIFRCNN)-based fusion method for remote sensing images. First, we put forward automatic learning matching module (ALMM). The utilization of ALMM can realize dynamic optimization matching of MS and PAN images with different resolutions. Second, we build feature fusion residual blocks suitable for RIFRCN to optimize network parameters. Finally, we input MS image and PAN image with different resolution and complete the fusion task through ALMM, fusion, and reconstruction module. Experiments on WorldView and QuickBird datasets manifest that the proposed RIFRCNN can effectively fuse MS and PAN images, and the fusion images generated in the experiments of true and false color images show better results than the state-of-the-art technology.

Our study demonstrates the potential of Airborne Visible/Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) hyperspectral data and radiative transfer modeling to retrieve the atmospheric carbon dioxide (CO₂) concentration from point source emission of one of India’s major coal-fed super thermal power plants at Kota, Rajasthan, India. AVIRIS-NG data with synchronous in situ measurements were collected as a part of the first ISRO-NASA joint airborne hyperspectral science campaign in India. The method utilized in our study includes theoretical simulations of at-sensor radiance in AVIRIS-NG spectral bands through the atmospheric radiative transfer model MODTRAN. Simulations were performed for the specific scene-sensor-atmospheric conditions pertaining to AVIRIS-NG overpass over Kota site. To eliminate the interfering effect of atmospheric water vapor with CO₂ concentrations, simulations pertaining to variable water vapor values varying from 0.25 to 4.5  g  /  cm² were also carried out, which in turn were utilized to normalize the water vapor effects. Based upon the simulation results pertaining to the specific absorption bands of CO₂ in the shortwave infrared spectral range of AVIRIS-NG radiance data, a theoretical model was established between continuum interpolated band ratio (CIBR) and CO₂ concentration. The CIBR  −  CO₂ model coefficients for each of the water vapor subranges were then applied to AVIRIS-NG L1 radiance data to obtain the CO₂ concentration map of the study area. A distinct CO₂ plume could be detected from the coal-fed Kota Super Thermal Power Station with the CO₂ concentration (X_CO2) of the order of more than 500 ppmv near the power plant stacks. A range of X_CO2 from 400 to 550 ppmv was observed within the scene. Our study has provided promising results in mapping the atmospheric CO₂ content from the point source using the high-resolution airborne hyperspectral AVIRIS-NG data.

Hyperspectral unmixing has become a key technology in hyperspectral image processing. The hyperspectral images contain rich spectral and spatial information. Most algorithms add various constraints and weight factors to make full use of this information and improve the unmixing accuracy. However, the improvement of unmixing accuracy often comes at the cost of large calculations. To solve this problem, we propose the superpixel and low-rank double-sparse regression unmixing algorithm, which can improve unmixing accuracy with lower computational complexity. This algorithm completes the unmixing process by performing two sparse regressions. The result of the first regression is added as a constraint to the second. For the first regression, we perform superpixel segmentation on hyperspectral image to exploit spatial information. To reduce computational complexity, we utilize the mean of each superpixel to represent all pixels in the superpixel for unmixing. In addition, we impose low rankness and sparsity constraints on the mean vector. For the second regression, we impose joint sparsity on the original hyperspectral image. In addition, we employ the reweighting strategy to further enhance the sparsity constraint on every sparse regression. Simulated and real data experiments have proven that the algorithm has improved the unmixing accuracy and unmixing time.

Remote sensing technology is an important means of ocean exploration. Unmanned aerial vehicle (UAV) remote sensing is widely used in small observation areas because of its flexible operation, timely data transmission, and high spatial resolution. According to these characteristics, we propose a ship attribute inversion algorithm based on the wake characteristics. First, a band selection method is used to reduce the data dimensionality and remove redundant information from UAV remote sensing data. Then, the ship wake is detected by extracting the shear wave (S-wave) characteristics. Finally, the ship’s motion parameters are inverted according to the S-wave attributes. The experimental results show that the method proposed can detect the wakes effectively and estimate ship velocity accurately using the UAV-obtained hyperspectral images.

The hyperspectral image (HSI) has the characteristics of high resolution, a large amount of data, and a high correlation of bands. In the many HSI processing technologies, image classification is the most basic one. Supervised classification is the most effective and common classification method. However, to achieve the ideal classification effect, supervised classification needs a large number of labeled samples, which requires a lot of time and labor. To solve this problem, we combine active learning (AL) and transfer learning and propose an iterative weighted framework based on active transfer learning. First, we solve the optimal reconstruction matrix and projection matrix by minimizing the reconstruction error. Then, we project labeled samples from the source and target domains into the common subspace. In the iteration of the common subspace, the classifier performance will be better with the increase of iteration number. In each iteration, the weighted strategy is adopted to weigh the samples of the source domain. In this way, valuable source domain labeled samples will get a larger weight, so as to help the classification of target domain samples better. At the same time, AL is used to screen out a certain number of samples of the target domain for manual labeling, which is added to the labeled samples set. Experiments on three data sets demonstrate the effectiveness and reliability of the proposed method.

Hyperspectral imagery denoising is a classical problem in hyperspectral remote sensing applications. In our previous work, we combined principal component analysis with wavelet shrinkage, and good denoising results were obtained. We combine minimum noise fraction (MNF) with video block matching and 3D filtering (VBM3D), which is a powerful video denoising method. After MNF transform, we automatically select k₀, the number of spectral band images as a threshold for denoising. We reduce the noise in spectral band images of MNF transformed data from k₀ to the last band image and do not denoise the first k₀  −  1 spectral band images. Finally, we perform an inverse MNF transform to obtain the denoised data cubes. We compare our MNF + VBM3D method with different denoising methods such as bivariate wavelet shrinkage (BivShrink), non-local means, SURELET, and block matching and 3D filtering (BM3D). Experimental results demonstrate that MNF + VBM3D achieves the best denoising results among almost all methods for three testing data cubes and with different noise levels.

Hyperspectral remote sensing data have been widely used in lithological identification and mapping. In existing studies, sufficient training samples are required. However, collecting sufficient labeled training samples for lithological classification in remote and inaccessible areas is generally time consuming, expensive, and even hard to implement, which causes the insufficient training sample problems in lithological classification using hyperspectral data. The semisupervised self-learning (SSL) method provides an alternative way of addressing this ill-posed problem by enlarging the training set from unlabeled samples using the limited labeled samples as a priori. This study evaluates and analyzes the SSL method in lithological mapping using EO-1 Hyperion hyperspectral image over a remote area. The performance of SSL with limited training samples was validated by comparing with multinomial logistic regression (MLR) and random forest (RF) classification using full training samples. The experimental results indicate that SSL method with limited training samples can produce results comparable with the MLR with all the training samples and better results than those of the RF with all the training samples. Therefore, the SSL method provides a useful way for hyperspectral lithological mapping with limited training samples over remote and inaccessible areas.

Vicarious calibration approaches using in-situ measurements are key techniques applied to on-orbit remote sensors. The in-situ measurements have advantages in terms of traceability and quantifiable accuracy that make them ideal for inter-consistency studies of on-orbit sensors. The Committee on Earth Observation Satellites Working Group on Calibration and Validation Infrared Visible Optical Sensors initiated the Radiometric Calibration Network (RadCalNet) to provide automated surface and top-of-atmosphere (TOA) reflectance data from multiple participating ground sites to the worldwide user community. RadCalNet standardizes protocols for collecting data, processes TOA reflectance, and provides International System of Units traceable uncertainty budgets for the five currently operating sites. RadCalNet data from the Railroad Valley Playa, Nevada, site was used to provide an SI-traceable, absolute radiometric calibration of the Enhanced MODIS Airborne Simulator as part of an overflight of opportunity during the FIREX-AQ campaign. The results shown here demonstrate the utility of RadCalNet data for absolute radiometric calibration of airborne imagers making flights of opportunity without the need for coordination with a field validation team.

Image stitching is an active research area in many photogrammetric and remote sensing applications. The main contribution of our work is to regard stitching as a multi-image matching problem that exploits advantages of invariant local features and homographic transformation to overcome photometric, geometric, and perspective distortions. Our proposed approach, referred to as Haar Invariant Feature with PROSAC, consists of three main steps. The first step extracts a set of independent keypoints by an adaptive local feature extractor based on Haar wavelet transform and scale invariant feature transform (SIFT). Then to increase the precision of matched points, we propose to substitute classical matching algorithms by a fast Haar-k-nearest neighbor algorithm and estimate the homography matrix using the Progressive Sample Consensus algorithm. Finally, we warp images according to the projective transformation and blend them with a Laplacian multi-band to create a perfect panorama. Experiments on various images show the effectiveness and the robustness of our approach compared to common image matching techniques such as SIFT, speeded up robust features, optimized random sample consensus with contrario, and image matching by affine simulation.

With the rapid development of optical remote sensing, it is urgent to find a reliable target detection method. Compared with traditional detection algorithms, a convolutional neural network has attracted considerable attention owing to its efficiency and high transitivity. However, different from general images, remote sensing images contain complex background information and dense small targets with changeable directions that make detection very challenging. To solve these problems and provide a stable and high-performance detection method, a rotated saliency fusion object detection (RSD) model based on “you only look once” (YOLO)v4 is established. First, salient image fusion technology is used to magnify target information. Second, the angle variable and rotated non-maximal suppression is introduced to improve the accuracy of rotated object detection by including the detection of dense objects. Third, the network structure is enhanced to improve the performance of small-target detection. Finally, the k-means algorithm and data enhancement are introduced to increase the robustness of the model. Extensive experiments demonstrate the superiority of the proposed model in detection speed and accuracy. The mean average precision of the proposed RSD model reaches 97.32% for the remote sensing images in a harbor area with an average detection speed of 13.41  s^−  1.

Laser scanning technology has enabled to study three-dimensional (3D) structures in forests. For example, airborne laser scanning (ALS) point cloud has been applied to detect individual trees and segment tree crowns. However, the accuracy of such approach remains a challenge because of the intersected crowns and complicated understories. We developed a metabolic theory-based algorithm for individual tree detection and crown segmentation from ALS data. The algorithm is composed of two parts, of which one is an unscaled transporting distance-based top-to-bottom detection approach, and the other is a scaled transporting distance-based segmentation approach. The unscaled transporting distance for detection is the absolute distance from tree root to crown and then to ALS point based on the tree structure models, whereas the scaled transporting distance for segmentation is the unscaled distance that is scaled by an initial tree height obtained during detection. This is based on a basic metabolic theory that vascular plants tend to minimize the material transporting distance from root to leaves. Hence, seven types of materials transporting distance models were built based on monopodial branching structure or crown-centered structure. The performance of the proposed approach was then further examined and compared with two typical canopy height model-based approaches and one typical point cloud-based approach, taking forest in Oxfordshire, UK, as a case study. The results showed that our approach can reach a recall of 1.00, a precision of 0.96, and an F-score of 0.98 and can reach to much higher accuracy for tree height (R²  =  0.8045) than the comparison approaches (R²  <  0.2) in the study plot. One of the main reasons that led to such low accuracy of comparison approaches is much overestimation of understory height with a mean error that is 2.9 times higher than that of our approach on average. Furthermore, ALS point-to-point level accuracy assessment shows 9.7% more ALS points were truly assigned in our approach than that of comparison approaches. It is noticed that the algorithm presented is not sensitive to the two key parameters: p (a percentage determining the threshold of unscaled transporting distances) ranging from 32.0% to 34.0% and λ (the proportion between the assumed crown center height and tree height) ranging from 0.70 to 0.90 based on our data set. Such high accuracy of our approach can greatly improve detections of individual tree and crown segmentation, especially in delineating understories in complex-structured forest.

A method of monitoring produce freshness with hyperspectral imaging and machine learning is described as a way to reduce food waste in grocery stores. The method relies on hyperspectral reflectance images in the visible–near-infrared spectral range from 387.12 to 1023.5 nm with a 2.12-nm spectral resolution. The images were recorded in a laboratory with the imager viewing produce samples illuminated by broadband halogen lights, but we also recorded and discussed the implications of the illumination spectrum of lights found in a variety of grocery stores. A convolutional neural network was used to perform freshness classification for potatoes, bananas, and green peppers. Additionally, a genetic algorithm (GA) was used to determine the wavelengths carrying the most useful information for age classification, with an eye toward a future multispectral imager. Hyperspectral images were processed to explore the use of RGB images, GA-selected multispectral images, and full-spectrum hyperspectral images. The GA-based feature selection method outperformed RGB images for all tested produce, outperformed hyperspectral imagery for bananas, and matched hyperspectral imagery performance for green peppers. This feature selection method is being used to develop a low-cost multispectral imager for use in monitoring produce in grocery stores.

Crop height is useful to monitor the effects of cropland reclamation. Unmanned aircraft system (UAS)-based observation through digital cameras has exhibited significant potential as a cost-effective way to predict crop height. However, the performance of UAS-based multispectral images has not been adequately analyzed in the context of estimating crop height. Therefore, our study investigates the performance of UAS-based multispectral images in terms of modeling and extracting crop height by considering a region of reclaimed cropland as an example. A framework is developed to model and extract the height of wheat and corn plants, the point cloud of varying densities is generated using the images, and geometric and physical outliers are removed from the point cloud. The flight altitude, threshold of extraction, and format of the data are examined to accurately describe and verify the canopy structure. The results indicate that UAS-based multispectral images can be used to estimate crop height. The relative accuracy of the estimated height of corn (<6  %  ) was higher than that of wheat (>9  %  ), but the opposite result was obtained in terms of absolute accuracy. The estimated crop height was significantly correlated with the measured above-ground biomass and can be used to monitor typical reclaimed cropland.

In remote sensing images, pixel-based classifiers use means or variance-covariance statistical parameters generated from training sample data sets. These parameters do not represent in totality about variations within the class. This research work enlightens each training sample’s role in handling heterogeneity within the class instead of using statistical parameters (mean). Modified possibilistic c-means fuzzy algorithm, capable of single class mapping, has been experimented to handle heterogeneity within the class. The mapping of mustard, wheat, and grass classes has been conducted using multispectral temporal images of Sentinel-2A/2B of Banasthali, Rajasthan region. It has been observed that while using individual samples in place of statistical parameters in fuzzy-based classifiers, the individual class identified has been least affected due to heterogeneity within the class. Mean membership difference for favorable and non-favorable classes as well as F-score, kappa, and overall accuracy have been calculated. Through all these parameters, individual samples as mean outperformed other training approaches.

Deep convolutional neural network semantic segmentation has played a significant role in remote sensing applications due to its capability of end-to-end training and automatic high-level features extraction. Highly accurate predictions have been recorded from different network architectures that are designed using the convolutional layers. However, the accuracy is mainly achieved at a very high cost of computation, which renders the network infeasible for real-time application on resource-constrained devices. Therefore, there is a need to establish a trade-off between accuracy and model efficiency. Our paper proposes a lightweight, highly accurate, memory-efficient segmentation network capable of deployment on resource-constrained devices. Our proposed network of 1.09 million trainable parameters attains an appreciable accuracy of 89.41% and 88.78% on the Vaihingen and Potsdam respective dataset of the ISPRS 2D Semantic Labeling Challenge. The result from the speed and the memory efficiency experiment shows that our proposed network is suitable for real-time remote sensing applications.

Semisupervised learning (SSL) techniques explore the progressive discovery of the hidden latent data structure by propagating supervised information on unlabeled data, which are thereafter used to reinforce learning. These schemes are beneficial in remote sensing, where thousands of new images are added every day, and manual labeling results are prohibitive. Our work introduces an ensemble-based semisupervised deep learning approach that initially takes a subset of labeled data D_l, which represents the latent structure of the data and progressively propagates labels automatically from an expanding set of unlabeled data D_u. The ensemble is a set of classifiers whose predictions are collated to derive a consolidated prediction. Only those data having a high-confidence prediction are considered as newly generated labels. The proposed approach was exhaustively validated on four public datasets, achieving appreciable results compared to the state-of-the-art methods in most of the evaluated configurations. For all datasets, the proposed approach achieved a classification F1-score and recall of up to 90%, on average. The SSL and recursive scheme also demonstrated an average gain of ∼2  %   at the last training stage in such large datasets.

In our study, a real-world application using the latest unmanned aerial vehicle (UAV) functionality is presented. Sugar beet is an important industrial crop in many countries. According to estimates, weeds in sugar beet fields dramatically reduce both the quantity and quality of sugar beet crops. Due to the spectral similarities between weeds and sugar beet seedlings, visual identification of weeds is extremely difficult in the sugar beet fields. In the present study, a lightweight, end-to-end trainable guided feature-based deep learning method, called DeepMultiFuse, has been developed to improve the weed segmentation performance using multispectral UAV images that aim to fulfill these requirements (to identify weed in sugar beet fields). The proposed architecture is composed of five basic concepts, including guided features, fusion module, dilation convolution operation, modified inception module, and gated encoder–decoder network extracting the object-level image representation for different scenes. The proposed network was trained on the generated dataset, including four multispectral orthomosaic reflectance maps using the RedEdge-M sensor in Rheinbach and three multispectral orthomosaic reflectance maps applying Sequoia sensor in Eschikon for mapping weed segmentation on the field. Experimental results demonstrated that the proposed network taking advantage of the feature fusion module-rich features outperforms state-of-the-art networks.

Classifying land cover using high-resolution remote-sensing images is challenging. The emergence of deep learning provides improved possibilities, but owing to the limitations of network structures, traditional convolutional neural network methods lose essential information about boundaries and small ground objects. We propose a superpixel-optimized convolutional neural network (SOCNN) framework to overcome this weakness. The SOCNN includes three modules: a semantic segmentation module, a superpixel optimization module, and a fusion module. The performance of the first module was evaluated using several common networks. PSPNet outperformed other networks, obtaining a pixel accuracy of 83.25%, a Kappa coefficient of 0.7862, and a mean intersection over union of 64.19%. Weighted loss was introduced to alleviate the effect of category imbalance, and the class pixel accuracy of category 11 improved by 19.77% with a weight of 20. The subpixel model was evaluated, and the pixel accuracy reached 83.43% with the superpixel-FCN method. Our superpixel optimized module improved the pixel accuracy of the object boundary by 1.37% when the fusion factor was 0.65. These results show that the SOCNN method is effective for recovering boundary information.

The unmanned aerial vehicle (UAV) plays an increasingly important role in monitoring and managing islands recently for their high feasibility and the miniaturization of sensors, which provide new possibilities for accurate island green cover mapping. We developed a framework that integrates UAV-acquired high-spatial resolution multispectral image and LiDAR data for effective object-based green cover mapping of Donkey Island in the Yellow Sea, China. LiDAR-derived structural and intensity information were combined with multispectral-derived spectral information for obtaining green cover objects. Five kinds of feature types [i.e., spectral, texture, height, intensity, and geometry features (GFs)] were calculated based on each object for green cover classification. Meanwhile, a multiple classifier system was adopted to improve the classification accuracy. The results indicate that the accuracy of green cover mapping could be significantly improved by the combination of multiple feature types. The inclusion of height and intensity features (IFs) can increase the overall classification accuracy by 7% and 5%, respectively, but the statistical significant differences are not found between these two feature types. The best green cover map is generated via a feature group obtained by the sequential backward selection with random forest method, reaching an overall accuracy of 88.5% and overall disagreement of 18.5%. Among the three major green cover classes, the accuracy of shrub class mapping improves the most when compared to classification using individual data, followed by tree and grass. Analysis of feature importance implies that spectral, height, and IFs are more beneficial to green cover mapping compared to texture and GFs. Furthermore, integrating multispectral and LiDAR data can provide more reliable green cover distribution maps and reduce the classification uncertainties.

Advanced differential SAR interferometry (A-DInSAR) has been used as an effective monitoring tool in mining environments. This work presents an investigation of surface displacement in the Germano iron mining complex (Mariana, Minas Gerais, Brazil), where on November 5, 2015, the Fundão tailings dam collapsed. Initially, the analysis was performed with 14 Sentinel-1 images, acquired from May 8, 2015, to November 4, 2015, and focused on the Fundão dam before its rupture to detect prior signs of surface deformations. The analysis after the Fundão dam collapse was conducted in the Germano area using 43 Sentinel-1 images, acquired from November 16, 2015 to May 15, 2017, and aimed at detecting surface displacement on the remaining infrastructure, the Sela&Tulipa and Selinha dikes and the main Germano tailings dam, which were directly or indirectly affected by the accident. The results reveal precursor signals of surface displacement at the Fundão dam prior to rupture, particularly on the left abutment setback sector where the collapse started according to a witness, reaching accumulated values of up to −60  mm. The results provided by the A-DInSAR analysis after the dam collapse indicate a stability in the Sela&Tulipa and Selinha dikes as well as in the main Germano dam, which is corroborated by in situ topographic measurements. On the auxiliary dikes within the Germano reservoir, accumulated displacement of up to −90  mm caused by settlements and continuous traffic of vehicles was detected. In addition, the detected accumulate displacement within the reservoir showed values of up to −120  mm, in general caused by settlements of the waste dry material due to the inactivity of the mining process. The investigation emphasized the important role that satellite-based A-DInSAR can play in mitigating possible accidents and for operational planning and risk assessment in mining operations.

Discrimination of crop varieties spanned over heterogeneous agriculture land is a vital application of polarimetric SAR images for agriculture monitoring and assessment. The covariance matrix of polarimetric SAR images is observed to follow a complex Wishart distribution for major classification tasks. It is true for homogeneous regions, but for heterogeneous regions, the covariance matrix follows a mixture of multiple Wishart distributions. We aim to improve the classification accuracy when the terrain under observation is heterogeneous. For this purpose, Wishart mixture model is employed along with expectation-maximization (EM) algorithm for parameter estimation. Elbow method helps us to devise the number of mixtures. The convergence of the EM algorithm depends on the choice of initial points. So, to improve the robustness of the model, different initialization approaches, such as random, K-means, and global K-means, are embedded in the EM algorithm. Further, the degrees of freedom is one of the crucial parameters of Wishart distribution. Therefore, the impact of different degrees of freedom is analyzed on classification accuracy. The method that is equipped with initialization technique along with optimum degrees of freedom is assessed using three full polarimetric SAR data sets of agriculture lands. The first two are benchmark data sets of Flevoland, Netherlands, region acquired by AIRSAR sensor, and third is our study area of Mysore, India, acquired by RADARSAT-2 sensor.

For the estimation of a dense disparity map from a rectified space-borne stereo image pair and generation of the most qualified digital surface model (DSM), new image matching algorithms are being developed. Our study has two main goals of the DSM quality validation of Worldview-4 (WV-4), which offers the highest ground sampling distance (GSD) (30 cm) of civilian optical space-borne missions together with Worldview-3 (WV-3), and performance comparison of semiglobal matching (SGM) and least squares matching (LSM), two of the most preferred image matching algorithms for space-borne data. In the Istanbul study area with a rough topographic structure, 1 m gridded DSMs were derived from geometrically corrected WV-4 stereo image pairs. The qualities were estimated by well-rounded visual and statistical approaches based on standard deviation and normalized median absolute deviation. In model-to-model comparisons, a very high-resolution airborne laser scanning (ALS) DSM was utilized as the reference model. The results demonstrated that the WV-4 DSM derived by SGM has the standard deviation of ±0.51  m (1.7 GSD) in height in open areas, whereas the LSM DSM has ±1.46  m. In addition, the visual quality of SGM is much better than LSM through building description potential.

Soil moisture is an important parameter that directly impacts crop productivity. Microwave signals are highly sensitive to soil dielectric constant and so they are used to derive soil moisture. The potential of entropy and alpha (H  /  α) decomposition for moisture estimations at 0- to 5-cm soil depth was assessed. The H  /  α parameters were extracted from the dual-polarimetric Sentinel-1 dataset. Also, we used the gray level co-occurrence matrix (GLCM) texture parameters extracted from Sentinel-1 and tested them for soil moisture retrieval. The generalized regression neural network (GRNN), neural network (NN), and support vector regression (SVR) algorithms were trained and tested for soil moisture estimation. Multiple input features, including Sentinel-1 intensities, GLCM parameters, and H  /  α parameters, were used for training these algorithms. For NN, three activation functions of rectified linear unit (ReLU), tanh, and sigmoid and for SVR three kernel functions of radial basis function (RBF), polynomial, and linear kernel were tested. The ReLU outperformed the other two activation functions with root mean squared error (RMSE) of 0.042  m³ m^−  3 and coefficient of determination (R²) of 0.72. For the SVR algorithm, the highest accuracies derived from the RBF kernel function with RMSE 0.053  m³ m^−  3 and R² of 0.51. Between all the three machine learning algorithms, the GRNN algorithm outperformed the other two algorithms with RMSE of 0.033  m³ m^−  3 and R² of 0.82. These results demonstrated the high potential of using polarimetric synthetic aperture radar data in combination with the machine learning algorithms for surface soil moisture monitoring.

Realtime stereo processing, particularly in computationally constrained environments such as the Mars 2020 Perseverance rover, requires image rectification to enable efficient one-dimensional correlation. A camera model has been designed to perform better image rectification for fish-eye lenses with a particular emphasis on the engineering cameras on Perseverance. The primary innovation is the use of a unit projection sphere rather than an image plane as the rectifying surface. For stereo rectification, the rows (for a horizontal pair) or the columns (for a vertical pair) of both cameras must lie on the same epipolar plane. Thus, we use a pair of moving planes to define the rows and columns. Each plane rotates about a dedicated static axis passing through the sphere center. Virtual pixels are located where the planes intersect with each other and the unit sphere.

The Jinghe River Basin is an important grain-producing area in northwest China; yet, serious soil erosion and water shortage problems have greatly limited the development of agriculture in the region. Evapotranspiration (ET) is a key component of rational water resources planning and water cycle mechanism exploration; yet, studies in this region have mainly focused on rainfall and runoff sand transport, neglecting the role of ET. Therefore, to understand the characteristics of ET in the Jinghe River basin, we estimated the ET during the vegetation growing season (April to October) in 2018 using the surface energy balance system (SEBS) based on remote sensing images and meteorological data, and analyzed the characteristics of its spatio-temporal distribution and its relationship with environmental factors. The results showed that (1) SEBS has good applicability in the study area, with R¹ between 0.57 and 0.77, RMSE between 0.98 and 1.15  mm  /  d, and MRE between 24% and 36% as verified with the lysimeter and ET products; (2) on the basin scale, the average daily ET showed a unimodal distribution with time, with variations ranging from 1.6 to 4.1 mm while spatially it showed that the mountainous forest and river catchment areas were overall higher than the loess plateau and hilly gully areas; (3) on the landscape scale, the vegetation with the highest average daily ET was broad-leaved forest, and the lowest was grassland. The vegetation with the highest ecological water demand was cultivated vegetation, and the lowest was meadow; (4) correlation analysis showed that rainfall, temperature, sunshine hours, and vegetation cover were positively correlated with ET. Among them, rainfall and temperature have the strongest correlation with ET and maybe the main factors affecting ET in the Jinghe River Basin.

The early identification and monitoring of landslides is of great significance to disaster assessment, mitigation, and prevention. Although interferometric radar technology can achieve large-scale and infinite-point with high-precision surface deformation monitoring, when the real deformation direction is different from the radar line of sight direction, its detection sensitivity will be reduced. We take the Xishan landslide group in Qinghai Province as the research object and use the ascending and descending Sentinel-1A data from January 9, 2019, to April 9, 2020, to reconstruct the slope direction velocity field based on conversion relationship; and using the mutual verification of ascending and descending orbit can obtain accurate and reliable unstable slope range. Adopting fusion of ascending and descending orbit data, with the decomposition of the deformation information of the Xishan landslide group, the vertical and east-west deformations of the slope are extracted, and the stability of the slope is analyzed based on the deformation characteristics of the landslide, topography, and site survey results.

Multi-source, multi-temporal, and multi-resolution satellite images have been increasing over a couple of years. The spatiotemporal data fusion technique could be a promising solution to achieve the advantage of temporal and spatial details of multi-platform remote sensing images. Due to the unavailability of required temporal resolution of normalized difference vegetation index (NDVI) of high-spatial-resolution dataset (e.g., Sentinel 2 A/B), the remote sensing-based crop monitoring is limiting in several parts of the world, primarily in the monsoon Asia and tropical region. We propose a deep learning-based framework to fill the gap of high-resolution NDVI value using corresponding low-spatial-resolution NDVI images. Time-series NDVI images from Sentinel 2 and Landsat 8 are used to implement the proposed framework. We evaluate our method on two study areas with different climatic conditions: subtropical and temperate. The results demonstrate that spatiotemporal data fusion with our proposed framework is very effective to synthesize and generate new datasets which represents the unique features of both input sources, namely, spatial and temporal details of each input image regardless of the climatic and temporal variations.

The fusion of a hyperspectral image (HSI) and a panchromatic image (PI) creates data that is beneficial to subsequent processing. Based on the traditional nonnegative matrix factorization (NMF)-based fusion method, the logarithmic hyperbolic cosine function is introduced as a spectral constraint. Under the constraint, the original HSI is decomposed into endmember and abundance matrices. The abundance matrix is then enhanced using PI, and the fused HSI is finally obtained with NMF reconstruction. In this way, the fused data have comparable spatial detail and spectral information. According to the experiments on the simulated and real data, the proposed fusion method obtains balanced fused results, and it is more suitable for the HSI fusion task.

Spectroscopy plays a vital role in the identification and characterization of minerals on terrestrial and planetary surfaces. We review the three different spectroscopic techniques for characterizing minerals on the Earth and lunar surfaces separately. Seven sedimentary and metamorphic terrestrial rock samples were analyzed with three field-based spectrometers, i.e., Raman, Fourier transform infrared (FTIR), and visible to near infrared and shortwave infrared (Vis–NIR–SWIR) spectrometers. Similarly, a review of work done by previous researchers on lunar rock samples was also carried out for their Raman, Vis–NIR–SWIR, and thermal (mid-infrared) spectral responses. It has been found in both the cases that the spectral information such as Si-O-Si stretching (polymorphs) in Raman spectra, identification of impurities, Christiansen and Restrahlen band center variation in mid-infrared spectra, location of elemental substitution, the content of iron, and shifting of the band center of diagnostic absorption features at 1 and 2  μm in reflectance spectra are contributing to the characterization and identification of terrestrial and lunar minerals. We show that quartz can be better characterized by considering silica polymorphs from Raman spectra, emission features in the range of 8 to 14  μm in FTIR spectra, and reflectance absorption features from Vis–NIR–SWIR spectra. KREEP materials from Apollo 12 and 14 samples are also better characterized using integrated spectroscopic studies. Integrated spectral responses felicitate comprehensive characterization and better identification of minerals. We suggest that Raman spectroscopy and visible and NIR-thermal spectroscopy are the best techniques to explore the Earth’s and lunar mineralogy.

Building damage assessment plays an essential role during post-disaster rescue operations. Given that labeled samples are difficult to timely obtain after a disaster, transfer learning attracts increasing attention. However, different sensors employed cause considerable discrepancies not only between historical and current scenes but also among historical scenes, which could exert an effect on transfer performance. Therefore, a multi-source adversarial domain adaptation (MADA) method is proposed in this paper to fulfill the task of post-disaster building assessment. This method consists of two phases. First, imageries of several historical scenes are transformed into the same style of the current scene through the CycleGAN model with a classifier, ensuring class invariance, to be fused to make an adapted source domain. Second, feature alignment between adapted source and target domains is executed based on adversarial discriminative domain adaptation. The MADA method enhances the transformed image quality, fully utilizes relevant information in historical scenes, solves inter-scene interference problems among historical images, and improves the transfer efficiency from historical to the current disaster scene. Two experiments are conducted with Hurricane Sandy, Irma, and Maria datasets as multi-source and target domains to validate MADA’s effectiveness. Results show that the classification performance is better than other methods.

Pan-sharpening is an effective method to obtain high-resolution multispectral (MS) images by fusing panchromatic (PAN) images with fine spatial structure and low-resolution MS images with rich spectral information. Many methods have been recently developed based on convolutional neural networks (CNNs) for pan-sharpening, which try to achieve better fusion performance using a large amount of training data. The huge demand for data of CNN-based methods results in huge pressure on transmission and storage conditions. Especially, PAN images occupy the main amount of data space because of whose spatial resolution is usually four times that of the corresponding MS images, which limits the application of pan-sharpening. We propose a CNN-based pan-sharpening method for compressed remote sensing to overcome this limitation, which can achieve pan-sharpening based on compressed data while eliminating compression artifacts. Specifically, we use compressed PAN images as the input and remove the compression artifacts, then fuse the PAN and MS images to achieve pan-sharpening. This method can greatly reduce the data capacity by data compression and ensure the pan-sharpening performance. Compared with the widely used state-of-the-art pan-sharpening approaches in a comprehensive evaluation, our method can obtain more promising results on compressed remote sensing images, and its superiority is thus demonstrated.

Bayesian methods are widely used to recover sparse signals from their incomplete measurements with Gaussian white noise. However, while recovering multidimensional signals, these methods are limited by their considerable computational complexities. To address this problem, we develop the truncated Bayesian recovery algorithm for compressive sensing (CS) of multidimensional signals, such as hyperspectral remote sensing images. For each dimension of multidimensional sparse signal, the entries are modeled by the type-II Laplacian prior, such that the support is indicated by the largest variance hyperparameters. The entries of noise tensor are assumed to be independent and Gaussian with zero mean. We approximately estimate the hyperparameters using the maximum a posteriori (MAP) process to determine the supports for each dimension. Based on the supports, the recovered sparse signal is obtained through the tensor-based least square strategy. The required number of iterative cycles can be significantly reduced, compared with conventional method that iterates the MAP process until convergence. Experimental results on synthetic data, hyperspectral images, and hyperspectral remote sensing images demonstrate that the proposed algorithm can obviously speed up multidimensional Bayesian CS recovery.

Accurate extraction of ground objects in mining areas using high spatial resolution imagery can effectively assist in the arrangement and deployment of the production work in mining areas. Therefore, it is essential to extract ground objects of mining areas quickly and accurately. However, the task is very difficult due to the variety of complex scenes, the high intra-class variation, low inter-class variation, similar appearance, and large size difference among the ground objects in mining areas. Traditional extraction methods based on RGB images can no longer meet the requirement of high precision. To solve these problems, we propose an attention-based multi-level feature fusion network (AMFNet), which extracts objects more accurately from unmanned aerial vehicle-based RGB images and digital surface model (DSM). The DSM of the mining area provides the height information of the ground objects, which is another supplementary key feature while the objects cannot be distinguished only by appearance in the RGB images. AMFNet is a dual-branch network of encoder–decoder structure using RGB images and DSM as the double-branch input separately. Specifically, the channel attention module, MF, and atrous convolution are introduced in the encoder network to make better use of dual-branch features. The decoder network upsamples the deep features and uses skip connections to fuse the multi-level shallow features obtained by the encoder network with the deep features. Finally, we set up multiple comparisons and ablation experiments in the test dataset, which demonstrate the performance advantages of AMFNet from different dimensions.

The unsupervised classification of hyperspectral images (HSIs) draws attention in the remote sensing community due to its inherent complexity and the lack of labeled data. Among unsupervised methods, sparse subspace clustering (SSC) achieves high clustering accuracy by constructing a sparse affinity matrix. However, SSC has limitations when clustering HSI images due to the number of spectral pixels. Specifically, the temporal complexity grows at a cubic ratio of the size of the data, making it inefficient for addressing HSI subspace clustering. We propose an efficient SSC-based method that significantly reduces the temporal and spatial computational complexity by splitting the HSI clustering task using similarity-constrained sampling. Our similarity-constrained sampling strategy considers both edge and superpixel information of the HSI to boost the clustering performance. This sampling strategy enables an intelligent selection of spectral signatures, and then, we split the clustering problem into multiples threads. Experimental results on widely used HSI datasets show that the efficiency of the proposed method outperforms baseline methods by up to 30% in overall accuracy and up to six times in computing time.

In recent years, Gabor filtering has been successfully applied in spectral–spatial hyperspectral image (HSI) classification tasks due to its strong power to characterize surface materials. Generally, the standard Gabor filter involves the real and imaginary parts. However, the traditional usage to integrate both the two parts, i.e., the Gabor magnitude feature, might weaken some unique characteristics of each part, therefore leading to possible information loss. To solve this problem, we proposed a flexible 3-D Gabor features fusion (F3DGF) approach to make better use of two parts of Gabor features, based on the phase-induced 3-D Gabor feature, which is rarely exploited before. As the term suggests, the phase-induced feature is guided by a phase parameter P, which can flexibly combine the information from two parts of Gabor components. The proposed F3DGF scheme explores all the phase-induced 3-D Gabor features by means of a decision-level fusion strategy, where the obtained probability outputs are directly gathered to generate the decision map. Experimental results on three real HSIs demonstrate that our approach exhibits good improvements, as compared to the traditional real part and the magnitude Gabor features-based classification methods. The results show great potential to introduce phase-induced 3-D Gabor features for classification tasks.

Recent technological advancements in satellite imagery have increased the production of remotely sensed images. Therefore, developing efficient methods for annotating these images has gained popularity. Most of the current state-of-the-art methods are based on supervised machine learning techniques. We propose a method called semisupervised multi-label annotizer (SS-MLA) that adapts vector-quantized temporal associative memory to annotate remotely sensed images. One of the advantages of SS-MLA over the supervised methods is that it extracts features not only from the given sample but also from similar samples that are previously seen without using an explicit attention mechanism. Thus SS-MLA enhances the learning efficiency of the training process. We conduct extensive performance comparisons with five different methods in the literature over four datasets. The comparison results indicate the success of the proposed method over the existing ones: SS-MLA generates the best results in 7 out of 11 comparisons.

We propose a panchromatic (Pan) and multispectral (MS) image fusion method in order to synthesize an MS image with enhanced spatial resolution and preserved spectral information of the original MS image. The proposed method is based on the spatial frequency and the à trous wavelet transform (SFATWT). The à trous wavelet transform (ATWT) is used to decompose Pan and MS images into low and high frequencies. The spatial frequency is exploited in the image fusion rules to improve the details or the high-frequency information which will be injected into the MS image. The ATWT reconstruction makes it possible to produce MS image with high spatial resolution. To validate the proposed method, two image sets are used. The first one is acquired by the SPOT (Satellite Pour l’Observation de la Terre) satellite with a resolution ratio of 2. The second one is the Algerian satellite AlSat 2A images with a resolution ratio of 4. The fused results are evaluated by means of qualitative and quantitative comparisons. The assessment of fused products has shown that the proposed method is promising.

Remote sensing (RS) oblique imagery provides valuable information of buildings’ facades. Facade detection using spectral-, spatial-, and texture-only features does not accurately separate facade and nonfacade regions in a single-view oblique image. Therefore, a facade index and a unimodal thresholding method were proposed to characterize and detect facade regions. This new index, named the probability-spatial facade index (PSFI), first highlighted facade areas. Then, the facade map was created through a histogram-based thresholding. In unimodal histograms, the thresholding procedure was according to the roots of the second derivative of a fourth-degree polynomial model that is fitted to the PSFI’s histogram. All the steps of the proposed method were implemented in the Google Earth Engine cloud computing platform and could automatically handle very high resolution oblique imagery (in terms of both angle and direction) without any limitations. Furthermore, various high- and low-rise buildings could be effectively processed without any assumptions about the structure of facades. The evaluation showed the high performance of the proposed method in different test areas, in which the average overall accuracies in distinguishing facade and nonfacade regions and in separating facade and rooftop regions were 97% and 86%, respectively.

Remote sensing images have the characteristics of a multi-scale, high resolution, and complex background, making the accuracy of small target detection too low. We propose an improved method based on the “you-only-look-once” v5 network model to make it more suitable for small target detection. First, we design a four-scale detection layer that improves the depth of the network and the accuracy of small target detection. Then, we introduce complete intersection over union-non-maximum suppression (NMS) to replace the commonly used NMS as a suppression of the prediction bounding box, and it also considers the overlap area, center distance, and aspect ratio to improve the accuracy of predicting bounding boxes. Furthermore, the remote sensing image is segmented into several fixed-size smaller images, making its size suitable for the input of the model to avoid target loss caused by directly scaling the size. The experimental results show that, compared to the four available detection methods, the improved algorithm significantly improves the detection accuracy of targets from small aircraft and ships in complex environments. The miss detection rate is also greatly reduced, especially for small targets, which verifies the effectiveness and robustness of the algorithm. Moreover, the detection rate has been greatly improved, which can meet the demand for real-time detection.

Given the unprecedented size of ultraspectral sounder data, there is a special process of radiance thinning in assimilating this data to reduce the data volume with minimal loss of atmospheric information. Considering the potential correlation between the selected data by radiance thinning and the unselected data, a lossless compression method for ultraspectral sounder data is proposed based on key information extraction and spatial–spectral prediction. Sensitive channels are first selected by stepwise iteration based on information entropy to maintain critical atmospheric information, and then auxiliary channels are further selected based on information content and correlation constraints to facilitate prediction. All of the selected channels are spatially thinned to generate key information, which is then used to predict original ultaspectral sounder data by spatially bicubic interpolation and spectrally sparse reconstruction. The residual errors are processed by the least-squares linear prediction to further reduce data redundancy. Together with the key information, the final residual errors are then fed into a range coder after positive mapping and histogram packing. Experimental results with IASI-L1C data show that the proposed method achieves an average compression ratio of 2.68, which is 4.7% higher than that of the typical methods, including JPEG-LS, JPEG-2000, M-CALIC, CCSDS-122.0, CCDS-123.0, and HEVC.

The detection of urban land use is of great significance to urban planning. With the development of deep learning technology, convolutional neural networks (CNN) are widely used in remote sensing image processing. However, existing CNN models lack the ability to fuse high-dimensional semantic information and low-dimensional position information of images and thus perform poorly in feature restoration. To obtain better urban land use detection results, the cross-dimensional feature fusion network (CDFFNet) is proposed in this work. In the CDFFNet, Class Feature Attention (CFA), cross-level feature fusion (CLFF), and Same Level Feature Gate Fusion (SLFGF) are designed. The CFA aggregates multi-scale and multi-receptive field input features to capture high-dimensional semantic information. The CLFF is designed to fuse features of different resolutions. It can extract and fuse semantic information of different dimensions, aggregate contextual information of features and distinguish the contextual relevance of each pixel. The SLFGF dynamically fuses high-dimensional semantic information and low-dimensional spatial information, so it could enhance information recovery of different categories. Compared with existing methods, the CDFFNet is significantly improved in accuracy. Its mean intersection over union on the Aerial Image Segmentation Dataset reaches 0.765.

In inverse synthetic aperture radar (ISAR) imaging, micro-Doppler (m-D) has always been a hot research topic. m-D will interfere with the imaging of the main body of the target, which will affect the image quality. However, m-D also contains micromotion information. In order to obtain a clearer image of the main body of the target and more micromotion information, some methods can be considered to separate the rigid body component and the micromotion component in the radar echo. To solve this problem, we propose a short-time chirplet decomposition algorithm to separate radar echo. This algorithm combines the window sliding mechanism with the adaptive chirplet decomposition algorithm. Compared with the adaptive chirplet decomposition algorithm and complex-valued empirical mode decomposition, it has a better separation effect and higher computational efficiency. The signal simulations, ISAR imaging simulations, and measured data experiments verify the effectiveness and superiority of the proposed algorithm.

Convolutional neural networks (CNNs) have achieved remarkable results in multispectral (MS) and panchromatic (PAN) image fusion (pansharpening) because of their strong image feature extraction ability. However, previous CNN-based pansharpening methods mostly use an ordinary convolution, which has a small receptive field in the convolution layer, has insufficient contextual information, and can only extract shallow features, which is not conducive to learning the complex nonlinear mapping relationship between the input image and the fused image. Therefore, this study proposes a pansharpening algorithm based on a multiscale densely convolutional neural network (MDCNN). First, a two-stream network is used for feature extraction, with two convolution layers to extract spectral information from MS images. The multiscale convolutional feature extraction module is designed to extract the spatial detail features of the PAN images. Second, the proposed multiscale densely connected modules and residual modules are used as the backbone of the fusion network. Finally, the deep features generated are reconstructed, and spectral mapping is used to retain spectral information to obtain a high-resolution fusion image. Experimental results using three satellite image datasets show that the proposed algorithm generates high-quality fusion images, and it outperforms most advanced pansharpening methods in subjective visual and objective evaluation indexes.

Spectral and spatial details abound in hyperspectral data. However, there are few labeled samples available, and obtaining them is expensive and time-consuming. As a result, improving the classification accuracy of HSI with limited training samples is a critical challenge. We propose a double-branch network with attention feature fusion (DBN-AFF) for HSI classification. One branch extracts spectral features, and the other extracts spatial features. It is capable of effectively preventing cross-branch interference. Furthermore, the AFF module is used instead of the channel concatenation method. It effectively fuses spectral and spatial features by considering the relationship between different channels. On the Indian Pines, Pavia University, and Salinas Valley datasets, DBN-AFF achieves 97.44%, 98.37%, and 97.60% on limited training samples, respectively.

Hyperspectral images contain rich spectral–spatial information. Therefore exploring spectral–spatial classifiers has become a mainstream trend in the field of hyperspectral image classification. However, current studies seldom delve into the contribution of the extracted spectral and spatial features to the subsequent classification task. To further explore the contribution of spectral and spatial features, we propose a classification framework based on capsule network (CapsNet). The framework consists of two branches, which extract spectral and spatial features to form capsules respectively. Subsequently, the spectral capsules and spatial capsules are sent to the dynamic routing layer to generate higher-level spectral–spatial capsules. In addition, we set up a primary–secondary relationship for the two branches, which indirectly reflects the contribution made by lower capsules forming higher capsules in the feature fusion stage. Our experiments, conducted on three widely used hyperspectral datasets and two sampling strategies, demonstrate that our proposed model had 4.60% to 8.80% improvements in OA compared with the original CapsNet. Compared to the capsule-based improved model (iCapsNet and Conv-Caps), our model also has a 1.20% to 2.50% improvement.

Remote sensing image processing has been widely used in environmental monitoring, terrain survey, military investigation, disaster early warning, and other fields. The remote sensing images matching is a key step, and the accuracy and real-time of the alignment have a great impact on these applications. Due to the high resolution and complexity of remote sensing images, scale-invariant feature transformation (SIFT) algorithm has the problems of high computational complexity and poor matching effect. Adaptive threshold adjustment algorithm has proposed in the remote sensing image matching, but minor changes in the contrast threshold can bring about drastic changes in the image matching quality, which will affect applications such as monitoring, measurement, and early warning. To improve the matching quality of a SIFT detector, an adaptive contrast threshold SIFT method based on image complexity calculation (CACT-SIFT) is proposed. The CACT-SIFT finds two contrast thresholds for the target image and the reference image, respectively, and achieves the target by minimizing the proposed criteria Cik. Experiments show that the method can be applied to the matching of remote sensing images. The information of reference images and target images can be detected at the same time, and the key points can be extracted in a robust way, with better accuracy and real-time accuracy of the alignment.

High-resolution multisource optical remote sensing images often have considerable non-linear radiation and scale differences that become more prominent as the image resolution increases. Since this problem may cause low registration accuracy, we propose an automatic registration algorithm based on point–line spatial geometric information (PLSGI). First, we propose an improved scale invariant feature transform algorithm using the quadtree uniformization, the Bhattacharyya distance, and the slope constraint to obtain many correct point pairs. Then we generate a new global line descriptor using PLSGI. In particular, the descriptor can resist the non-linear radiation and scale differences of multisource remote sensing images. Finally, we use a piecewise linear model to warp the sensed images based on a set of tie points that consist of corresponding features and corresponding intersections. The effectiveness of the proposed algorithm was validated using multiple groups of high-resolution remote sensing images. The experimental results indicated that our algorithm is highly generalized and robust for high-resolution multisource remotely sensed images.

Waveform design is a key component of cognitive radar. However, waveforms designed using a single criterion have difficulty meeting the multi-working mode of radar. We present a joint waveform design method in the presence of signal-dependent interference based on a long short term memory network. The proposed method combines characteristics of mutual information (MI) and signal to interference-plus-noise-ratio (SINR) criteria and achieves a smooth trade-off between them. The minimum mean square error (MMSE) method is set as a comparison to prove the enhanced capabilities in detection and recognition. Here, we initiate a final recognition rate indicator of radar to measure the radar’s comprehensive performance. As a result, compared with the MI criterion, SINR criterion, and MMSE method, to use as a transmission signal, the proposed paradigm increased the final recognition rate by 4.08%, 6.89%, and 5.25%, respectively, verifying the effectiveness of the proposed method.

The lack of shortwave (SW, visible, and near-infrared) geostationary satellite data at night results in degradation of many weather forecasts and real-time diagnostic products. We present a method to extrapolate SW GOES-16 advanced baseline imager data through night using nighttime longwave (LW, infrared) observations and the relationships between LW and SW data observed during the previous day. The method is not a forecast since it requires LW nighttime observations but can provide continuity through day, night, and satellite terminator hours. To provide performance statistics, the algorithm is applied during the day so the SW extrapolations can be compared to observations. Typical mean absolute errors (MAEs) range from 1.0% to 12.7% reflectance depending on the SW channel. These MAEs can be predicted using a diagnostic metric called 0-h MAE which quantifies the quality of the algorithm’s input data. In addition to quantitative error statistics, three case studies are presented, including an animation of extrapolated imagery from dusk through dawn. Considerations for future improvements include use of convolutional neural networks and/or object-based extrapolations where mesoscale features are extrapolated individually.

Aboveground carbon (AGC) maps will assist with the monitoring and verification of carbon stored through restoration of subtropical thicket. Field methods are required to capture AGC ground truth data at the plot scale, but they are highly impractical for large area mapping of carbon stocks. Against this background, a remote sensing method to estimate AGC from very high spatial resolution multi-spectral imagery was developed. AGC ground truth was acquired for 85 plots in a small study area in the Baviaanskloof, South Africa. Using the ground truth, univariate and multivariate snapshot models were developed to predict AGC from features in a 2017 WorldView-3 (0.34-m resolution 8-band) satellite image. Informative features were selected from a large set of spectral, textural, and vegetation index features using stepwise forward selection. Using this approach, the multivariate model produced a coefficient of determination (R²) of 0.886 and root mean square error of 2.862  t C ha^−  1. This study demonstrates the efficacy of regression approaches for estimating AGC from multi-spectral imagery and provides a foundation for the spatial and temporal extension of AGC remote sensing in the thicket biome.

Airborne lidar data for fishery surveys often do not contain physics-based features that can be used to identify fish; consequently, the fish must be manually identified, which is a time-consuming process. To reduce the time required to identify fish, supervised machine learning was successfully applied to lidar data from fishery surveys to automate the process of identifying regions with a high probability of containing fish. Using data from Yellowstone Lake and the Gulf of Mexico, multiple experiments were run to simulate real-world scenarios. Although the human cannot be fully removed from the loop, the amount of data that would require manual inspection was reduced by 61.14% and 26.8% in the Yellowstone Lake and Gulf of Mexico datasets, respectively.

Domain adaptation is a technology enabling aided target recognition and other algorithms for environments and targets with data or labeled data that is scarce. Recent advances in unsupervised domain adaptation have demonstrated excellent performance but only when the domain shift is relatively small. We proposed targeted adversarial discriminative domain adaptation (T-ADDA), a semi-supervised domain adaptation method that extends the ADDA framework. By providing at least one labeled target image per class, used as a cue to guide the adaption, T-ADDA significantly boosts the performance of ADDA and is applicable to the challenging scenario in which the sets of targets in the source and target domains are not the same. The efficacy of T-ADDA is demonstrated by cross-domain, cross-sensor, and cross-target experiments using the common digits datasets and several aerial image datasets. Results demonstrate an average increase of 15% improvement with T-ADDA over ADDA using just a few labeled images when adapting to a small domain shift and afforded a 60% improvement when adapting to large domain shifts.

Determining the amount of grain stored in silos is very important for accurate commercial inventory planning. A convolutional neural network (CNN) is developed for the first time to determine the amount of the grain using step-frequency continuous wave radar (SFCWR) signals. The radar reflection signal of different grain quantity for different grain surface patterns is gathered by means of a constructed experimental setup. 5681 measurements are performed in the scaled model silo containing different weights (0 to 20 kg) grain stacked as different surface patterns. The dataset is then created using the spectrograms of SFCWR signals. While 1420 data randomly selected from the dataset are used for testing, the remaining 4261 data are used for training. The results are then compared with the pretrained CNNs, demonstrating the superiority of the proposed method. The accuracy of the methods is given with metric parameters for both classification and regression. The proposed multitask CNN model obtained higher performance with 0.2865 MAE, 0.5053 MAPE, 0.8047 MSE, and 0.8971 RMSE for regression task and 99.23% accuracy, 99.09% sensitivity, 99.52% specitivity, 99.42% precision, 99.25% F1-score, and 98.83% MCC for classification. These metric performances are better than the previous study with 3.29 MAPE in the literature. The results obtained reveal that, with proper modeling and successful training, CNNs can be effectively used for the quantity measurement applications of the grain stacks.

The Conne River watershed is dominated by wetlands that provide valuable ecosystem services, including contributing to the survivability and propagation of Atlantic salmon, an important subsistence species that has shown a dramatic decline over the past 30 years. To better understand and improve the management of the watershed, and in turn, the Atlantic salmon, a wetland inventory of the area is developed using advanced remote sensing methods including field-collected data, object-based image analysis of Sentinel-1, Sentinel-2, and digital elevation model Earth observation data. The resulting classification maps consisted of bog, fen, swamp, marsh, and open water wetlands with an overall accuracy of 92% and a kappa coefficient of 0.916. Among wetland classes, user and producer accuracies range between 84% and 100%. Results show the dominance of peatland wetlands such as bog and fen, and the relative rareness of marsh wetlands.

Anomaly detection (AD) is a crucial task for detecting salient objects in cluttered backgrounds. Classical AD algorithms based on statistical models or geometric models have achieved acceptable detection results. However, most of them do not hierarchically extract deep features or consider the spatial structure of the images. We propose a method that combines the reconstruction error of autoencoder (AE) and spatial morphological characteristics to estimate anomalousness. The reconstruction errors and the spatially dominant information are comprehensively considered. Specifically, given the compression capability of the AE, we use the dimensionality reduction results obtained by encoding for analyzing local pixel difference; an adaptive dual window is employed in this process. The morphological transformation commonly used in edge detection is utilized to refine the small space anomalies. Experimental results on different hyperspectral images show that the proposed AE and spatial morphology extraction-based approach significantly surpasses several traditional alternatives.

Coastal tidal wetlands are significant and vulnerable aquatic ecosystems. Quantitative remote sensing of biophysical and biochemical properties in coastal tidal wetland habitats through the inversion of physical models has vital practical significance for monitoring ecosystem function, environmental restoration, the global carbon, and nitrogen cycles. The objectives of this research were to map leaf area index (LAI), chlorophyll content, as well as the sensible heat flux, latent heat flux, and productivity in Chongming Dongtan coastal tidal wetland habitats of China and to provide a mapping protocol of biophysical and biochemical properties in Chongming Dongtan wetland for environmental protection and restoration as well as assessment and monitoring. In order to obtain significant information for biodiversity protection and management, a method based on the Soil Canopy Observation of Photosynthesis and Energy fluxes model and lookup table approach has been developed. The results derived from our study contain the reflectance values of Scirpus mariqueter and Phragmites australis spectra are lower than those commonly found for vegetated areas. This may be caused by the dark soil background and low LAI. Moreover, as for the values of latent heat flux (between 120 and 190  W m^−  2) and productivity (between 3 and 14  μmol m^−  2 s^−  1), mudflats are lower than vegetation. However, the values of sensible heat flux (between 140 and 170  W m^−  2) of mudflats are higher than that of vegetation.

Recently, deep learning models based on convolutional neural networks (CNN) remain dominant in hyperspectral image (HSI) classification. However, there are some problems in CNN models, such as not good at modeling the long-distance dependencies and obtaining global context information. Different from the existing CNN-based models, an innovative classification method based on the transformer model is proposed to further improve the classification accuracy of HSI. Specifically, the proposed method first extracts the extended morphological profile (EMP) features of HSI to make full use of the spatial and spectral information while effectively reducing the number of bands. Next, a deep network model is constructed by introducing the transformer-iN-transformer (TNT) modules to carry out end-to-end classification. The outer and inner transformer models in the TNT module can extract the patch-level and pixel-level features, respectively, to make full use of the global and local information in the input EMP cubes. Experimental results on three public HSI data sets show that the proposed method can achieve better classification performance than the existing CNN-based models. In addition, using the transformer-based deep model without convolution to classify HSI provides a new idea for related research.

The erratum corrects an error in the originally published version of the article.