Search

A temporal analysis of landscape change patterns and consequent evaluation of urban expansion has been studied for the Dera Ghazi Khan city which is developing fast since 2000 as an emerging metropolis of South Punjab. The geospatial techniques particularly remote sensing augmented with GIS are vital tools and are mostly used in studying landscape changes and decision making for sustainable urban development. Four Landsat images (1991, 2001, 2011 and 2021) obtained from the free web of GLFC have been analysed. A supervised classification algorithm was applied to all images and results have been presented in both tabular and graphic forms. Present study, focuses upon two land-use classes i.e. built up and open areas, which will be monitored by land-use patterns of Dera Ghazi Khan city during 1991 to 2021 using satellite images. The graphical depiction of all the landscape changes, which have taken place in D. G. Khan city in the form of tables and maps provides detailed information about the degree and ratio of urban expansion and the changes in land use that occurred during the stipulated period.The classification was based on four categories i.e., urban land, bare soil, vegetation, and water; while maps were prepared to present the temporal changes in D. G. Khan city. A total of 52 sq./kms areas was selected for this study designated as the Dera Ghazi Khan city. According to the results, 33% of the urban area has been expanded in Dera Ghazi Khan from 1991 to 2021 and 19% vegetation cover has been reduced. It is a very alarming situation that agricultural land is decreasing because of urban expansion. The results revealed the significance of remote sensing and GIS in evaluating landscape changes.

With vast usage of Spatio-temporal data usage in various domains and capturing data as remote sensing, GPS, and ST data. STDM is critical for various domains such as transportation, public safety, medical and environmental statistics. In this article, ST characteristics are explained in seven different types and followed by models for STDM in different domains including medical, mobile usage, sensors, industrial area, weather analysis, data querying, traffic control. Finally, the research and future direction.

Ephemeral gullies are widely distributed in the hilly and gully region of the Loess Plateau and play a unique role in the slope gully erosion system. Rapid and accurate identification of ephemeral gullies impacts the distribution law and development trend of soil erosion on the Loess Plateau. Deep learning algorithms can quickly and accurately process large data samples that recognize ephemeral gullies from remote sensing images. Here, we investigated ephemeral gullies in the Zhoutungou watershed in the hilly and gully region of the Loess Plateau in China using satellite and unmanned aerial vehicle images and combined a deep learning image semantic segmentation model to realize automatic recognition and feature extraction. Using Accuracy, Precision, Recall, F1value, and AUC, we compared the ephemeral gully recognition results and accuracy evaluation of U-Net, R2U-Net, and SegNet image semantic segmentation models. The SegNet model was ranked first, followed by the R2U-Net and U-Net models, for ephemeral gully recognition in the hilly and gully region of the Loess Plateau. The ephemeral gully length and width between predicted and measured values had RMSE values of 6.78 m and 0.50 m, respectively, indicating that the model has an excellent recognition effect. This study identified a fast and accurate method for ephemeral gully recognition in the hilly and gully region of the Loess Plateau based on remote sensing images to provide an academic reference and practical guidance for soil erosion monitoring and slope and gully management in the Loess Plateau region.

In recent years, parallel to the developments in satellite technology, obtaining and processing remote sensing images has become quite common. While airports are the first points to be targeted by enemy forces in times of war, they are very critical points in times of peace due to their significance for transportation, trade, and economy networks. The runways are the most distinctive feature of airports. There are many studies on detecting the runways in remote sensing images (RSIs). However, existing methods for detecting the runway objects that have an excessive width in high-resolution (4137 x 4552 pixels and above) RSIs may be insufficient. In this study, a Divide and Conquer Object Detection (DACOD) method is proposed for the runway objects that have an excessive width in high-resolution RSIs. In the proposed method, images are divided into images of 1024 x 1024 pixels, and the runway objects in these images are detected as oriented. Then, the detection results are merged by using the angles and the final runway detection results are obtained. The experimental results demonstrate that the proposed model yields good results (%81.5 mAP). This is an 11% mAP increase when compared to the best results in The State of The Art (SOTA) object detection models using the same dataset.

In Bangladesh, riverbank erosion has turned into one of the most frequently occurring natural disasters, and it affects the socio-economic and livelihood status as well as the environment of the adjoining areas. Kirtankhola River in Barishal, is this kind of river. This study assesses the amount of erosion-accretion, examines river profile, detecting key anthropogenic factors behind it, and signifying channel shifting of the river within 10 years (from 2009 to 2019). The study used Landsat Satellite imageries, geospatial techniques, and a semi-structured questionnaire survey throughout the study area. The study showed that erosion dominants at the left bank of river (17.1 ha/year), whereas the accretion rate is higher at the right (19.72 ha/year). Besides, cross-section at selected points of the river providing an evident indication of the river widening thus river course shifting. According to 61% of the participants in the questionnaire, high waves generated through water vehicles are the main culprit for this extensive erosion. The number of launches has increased significantly within the studied time-period, and most of the launches run over 30 km/h velocity, where it should be less than 20km/h. The study explored a mechanism of how centrifugal force produced by water vehicles increases the intensity of riverbank erosion in the meandering area.

The Ganges-Jamuna-Padma confluence is one of the world's most active confluences. The confluence of two of the world's greatest rivers, the Ganges and the Brahmaputra, makes this a globally significant site. Severe erosion along the banks has been caused by morphological changes in this region. Riverbank erosion is one of Bangladesh's most serious problems, as it necessitates costly intervention. Riverbank erosion in Bangladesh affects millions of people each year as a result of erosion in this confluence zone. As a result, it's critical to comprehend the confluence's morphological changing pattern. This study aims to quantify actual bank shifting around the confluence of the Ganges, Jamuna, and Padma in terms of shifting rate and area during a twenty-five-year period (1990-2015). To conduct this study the collected satellite image were geo-referenced and digitize bank lines from using ArcGIS program. The bank line is the linear structure that divides the river channel's outer border from the flood plains. The distance between the extreme margins of the left and right banks, including mid-channel sandbars, was measured to determine channel width variation. To assess the maturity of change, this time frame is subdivided into five phases, each lasting five years. In addition, the long-term shift from 1972 to 2015 is qualitatively noticeable. This morphological alteration was studied using LANDSAT satellite images. The research gives current and trustworthy information on the Ganga-Jamuna confluence's planform dynamics. This research will be useful in the planning and execution of drainage development plans and erosion control strategies in this critical confluence zone.

The study involves an object-based segmentation method to extract feature changes in tropical rainforest cover using Landsat image and airborne LiDAR (ALS). Disturbance event that are represents the changes are examined by the classification of multisensor data; that is a highly accurate ALS with different resolutions of multispectral Landsat image. Disturbance Index (DI) derived from Tasseled Cap Transformation, Normalized Difference Vegetation Index (NDVI), and the ALS height are the variables for object-based segmentation process. The classification is categorized into two classes; disturbed and non-disturbed forest cover using Nearest Neighbor (NN), Random Forest (RF) and Support Vector Machine (SVM). The overall accuracy ranging from 88% to 96% and kappa ranging from 0.79 to 0.91. Mcnemar’s test p-value (

Due to the attractive potential in avoiding the elaborate definition of anchor attributes, anchor-free-based deep learning approaches are promising for object detection in remote sensing imagery. CornerNet is one of the most representative methods in anchor-free-based deep learning approaches. However, it can be observed distinctly from the visual inspection that the CornerNet is limited in grouping keypoints, which significantly impacts the detection performance. To address the above problem, a novel and effective approach, called GroupNet, is presented in this paper, which adaptively groups corner specific to the objects based on corner embedding vector and corner grouping network. Compared with the CornerNet, the proposed approach is more effective in learning the semantic relationship between corners and improving remarkably the detection performance. On NWPU dataset, experiments demonstrate that our GroupNet not only outperforms the CornerNet with an AP of 12.8%, but also achieves comparable performance to considerable approaches with 83.4% AP.

Urban sprawl is a phenomenon with a negative effect which, over time, consumes the living environment . The aim of this research is to analyze and evaluate the urban sprawl of Annaba situated at the northeastern part of Algeria in a diachronic approach. The studied area contains twelve agglomerations. Through the supervised classification of a time series of data provided by multi spectral sensors mounted on LANDSAT 8 Operati onal Land Imager (OLI), LANDSAT 7 Thematic Mapper Plus satellite platforms (ETM +) and L ANDSAT 45 Thematic Mapper (TM) the urban sprawl was examined in different periods of 1987, 2013 and 2017, on different scales. This study displayed that a retrospective understanding of the timing and magnitude of Annaba sprawl in a spatio temporal dimension allows updating the phenomenon to better manage it in the future on either macro or micro scale as a decision support tool.

The use of remote sensing for forest health monitoring has increased in popularity over the years, with improved quality in spatial and spectral resolutions. However, satellite revisit times are ge...

Deep learning models have become the mainstream algorithm for processing computer vision tasks. In object detection tasks, the detection box is usually set as a rectangular box aligned with the coordinate axis, so as to achieve the complete package of the object. However, when facing some objects with large aspect ratio and angle, the bounding box has to become large, which makes the bounding box contain a large amount of useless background information. In this study, a different approach is taken, using a method based on YOLOv5, the angle information dimension is increased in head part and angle regression added at the same time of the border regression, combining ciou and smoothl1 to calculate the bounding box loss, so that the resulting border box fits the actual object more closely. At the same time, the original dataset tags are also preprocessed to calculate the angle information of interest. The purpose of these improvements is to realize object detection with angles in remote-sensing images, especially for objects with large aspect ratios, such as ships, airplanes, and automobiles. Compared with the traditional object detection model based on deep learning, experimental results show that the proposed method has a unique effect in detecting rotating objects.

This study investigates the different aspects of multimedia computing in Video Synthetic Aperture Radar (Video-SAR) as a new mode of radar imaging for real-time remote sensing and surveillance. This research also considers new suggestions in the systematic design, research taxonomy, and future trends of radar data processing. Despite the conventional modes of SAR imaging, Video-SAR can generate video sequences to obtain online monitoring and green surveillance throughout the day and night (regardless of light sources) in all weathers. First, an introduction to Video-SAR is presented. Then, some specific properties of this imaging mode are reviewed. Particularly, this research covers one of the most important aspects of the Video-SAR systems, namely, the systematic design requirements, and also some new types of visual distortions which are different from the distortions, artifacts and noises observed in the conventional imaging radar. In addition, some topics on the general features and high-performance computing of Video-SAR towards radar communications through Unmanned Aerial Vehicle (UAV) platforms, Internet of Multimedia Things (IoMT), Video-SAR data processing issues, and real-world applications are investigated.

Evapotranspiration (ET) is one of the main fluxes in the global water cycle. As the Pampa biome carries a rich biodiversity, accurate information on the ET dynamics is essential to support its proper monitoring and establish conservation strategies. In this context, we assessed an operational methodology based on the S-SEBI model to estimate energy fluxes over the natural grasslands of the Brazilian Pampa between 2014-2019. The S-SEBI is a remote sensing-based ET model that requires a minimum of meteorological inputs, being fully applicable for regions lacking in situ observations. Therefore, we investigated the model performance considering radiation data from both ERA5 reanalysis and Eddy Covariance measurements. Furthermore, comparisons from satellite-based estimates with in situ data were performed with and without energy balance closure (EBC). Results showed that the RMSE is impacted in 0.32 mm/day when reanalysis data are used, indicating that the meteorological inputs have low sensitivity on daily ET estimates from the S-SEBI model. In contrast, the instantaneous energy balance components are more affected. The strong climate seasonality of the Brazilian Pampa impacts the evaporative fraction, which is more evident in late summer and autumn and may compromise the performance of the model. The effects in the daily ET are lower when in situ data without EBC are considered as ground truth. However, they are less correlated with the remote sensing-based estimates. These insights are useful to monitor water and energy fluxes from local to regional scale and provide the opportunity to capture ET trends over the biome.

Remote sensing image scene classification has attracted great attention because of its wide applications. Although convolutional neural network (CNN)-based methods for scene classification have achieved excellent results, the large-scale variation of the features and objects in remote sensing images limits the further improvement of the classification performance. To address this issue, we present multiscale representation for scene classification, which is realized by a global-local two-stream architecture. This architecture has two branches of the global stream and local stream, which can individually extract the global features and local features from the whole image and the most important area. In order to locate the most important area in the whole image using only image-level labels, a weakly supervised key area detection strategy of structured key area localization (SKAL) is specially designed to connect the above two streams. To verify the effectiveness of the proposed SKAL-based two-stream architecture, we conduct comparative experiments based on three widely used CNN models, including AlexNet, GoogleNet, and ResNet18, on four public remote sensing image scene classification data sets, and achieve the state-of-the-art results on all the four data sets. Our codes are provided in https://github.com/hw2hwei/SKAL.

Remote sensing based various land surface and bio-physical variables like Normalized Difference Vegetation Index (NDVI), Land Surface Temperature (LST), surface albedo, transmittance and surface emissivity are useful for the estimation of spatio-temporal variations in evapotranspiration (ET) using Surface Energy Balance Algorithm for Land (SEBAL) method. These variables were estimated under the present study for Ozat-II canal command in Junagadh district, Gujarat, India, using Landsat-7 and Landsat-8 images of summer season of years 2014 and 2015. The derived parameters were used in SEBAL to estimate the Actual Evapotranspiration (AET) of groundnut and sesame crops. The lower values NDVI observed during initial (March) and end (May) stages of crop growth indicated low vegetation cover during these periods. With full canopy coverage of the crops, higher value of NDVI (0.90) was observed during the mid-crop growth stage. The remote sensing-based LST was lower for agricultural areas and the area near banks of the canal and Ozat River, while higher surface temperatures were observed for rural settlements, road and areas with exposed dry soil. The maximum surface temperatures in the cropland were observed as 311.0 K during March 25, 2014 and 315.8 K during May 31, 2015. The AET of summer groundnut increased from 3.75 to 7.38 mm.day-1, and then decreased to 3.99 mm.day-1 towards the end stage of crop growth. The daily AET of summer sesame ranged from 1.06 to 7.72 mm.day-1 over different crop growth stages. The seasonal AET of groundnut and sesame worked out to 358.19 mm and 346.31 mm, respectively. The estimated AET would be helpful to schedule irrigation in the large canal command.

Remote sensing image segmentation provides technical support for decision making in many areas of environmental resource management. But, the quality of the remote sensing images obtained from different channels can vary considerably, and manually labeling a mass amount of image data is too expensive and inefficiently. In this paper, we propose a point density force field clustering (PDFC) process. According to the spectral information from different ground objects, remote sensing superpixel points are divided into core and edge data points. The differences in the densities of core data points are used to form the local peak. The center of the initial cluster can be determined by the weighted density and position of the local peak. An iterative nebular clustering process is used to obtain the result, and a proposed new objective function is used to optimize the model parameters automatically to obtain the global optimal clustering solution. The proposed algorithm can cluster the area of different ground objects in remote sensing images automatically, and these categories are then labeled by humans simply.

Groundwater upwellings provide warmer, stable overwinter temperatures for developing salmon embryos, which may be particularly important in cold, braided, gravel-bed sub-Arctic rivers. We used a three-year time series of aerial counts and remote sensing to estimate the distribution of low and high aggregations of spawning fall chum salmon (Oncorhynchus keta), classify approximately 0.5 km long river segments by geomorphic channel type, and map thermal variability along a 25.4 km stretch of the Teedriinjik River, Alaska. We used a dynamic multistate occupancy model to estimate detectability, occupancy, and the dynamics of spawning aggregations among river segments. Detectability was higher for large (>150) relative to smaller aggregations. Unoccupied segments were likely to remain so from year to year; low abundance spawning segments were dynamic and rarely remained in that state for multiple years, while ∼20%–35% of high abundance segments remained stable, indicating the presence of high-quality spawning habitat. Spawning habitat use was associated with warmer water temperatures likely caused by groundwater upwellings. We identified spawning habitat characteristics and trends in usage by fall chum salmon, which will inform land management decisions and assist in evaluating impacts of shifting climate conditions and resource management on Arctic salmon populations.

At present, due to the limitations of satellite launch cost and existing technology, it is scarcely possible to obtain single remotely sensed images with both fine spatial resolution and high temporal resolution at the same time freely. For solving this kind of predicament, an effective method is to fuse multi-source remote sensing data by using spatial-temporal super-resolution mapping (STSRM) algorithms. STSRM is developed on the foundation of super-resolution mapping (SRM), which is used for generating land cover map with finer spatial resolution by allocating sub-pixels position in the mixed pixels of coarse remotely sensed images. This review summarizes the existing mainstream models of spatio-temporal super-resolution mapping and concludes the advantages and limitations of these methods. At the same time, this paper analyzes methods of classification accuracy assessment, expounds the existing problems and challenges, and makes a forward-looking prospect for the future development direction of spatio-temporal super-resolution mapping.

Land use and land cover classification from a remote sensing image is a long standing research problem. It ranges from simple classifications like mapping water bodies to complex classifications like crop and forest strands. Crop image classification is complex because of various stages of growth of the same crop, same spectral values for various crops, an other multitude of problems. Crop image classification is very essential for agriculture monitoring, crop yield production, global food security, etc. A new unsupervised algorithm, Spherical Contact Distribution Classification Algorithm (SCDCA) is proposed in this paper which uses mathematical morphology, spherical contact distributions, and first order statistics. Later SCDCA is compared with linear contact distribution classification algorithm (LCDCA). Quantitative analyses prove the efficiency of the algorithm and present that the complexity of SCDCA is very much less when compared to that of LCDCA.

Land use and land cover changes have diverse envirnomental implications. An increasing number of studies are showing a direct link between land cover change and the acceleration of climate change. Digital classification of remotely sensed images is a strong tool for assessing the degree of environmental transformation due to anthropogenic factors. The aim of this work is to use remote sensing and digital image classification to obtain cartographies of land use and land cover in the Segura Hydrographic Demarcation, for the period between 1986 and 2019 and to analyse the observed changes. Random Forest with Landsat 5, 7 and 8 images was used to achieve this goal; although the overall accuracy achieved indicates that there is room for improvement in the model fit, the classifications obtained are considered reliable.in addition, they will be analysed to determine the most significant changes that the land surface of the study area has undergone. With regard to land use and land cover, a decrease in forest use and an increase in agricultural use, the area covered by scrubland, bare soil and greenhouses are obtained. Los cambios en los usos y coberturas del suelo conllevan consecuencias ambientales de diverso tipo. Cada vez son más los estudios que demuestran la existencia de una relación directa entre los cambios en la cobertura terrestre y la aceleración del cambio climático. La clasificación digital de imágenes obtenidas mediante teledetección es una potente herramienta para evaluar el grado de transformación ambiental debido a factores antropogénicos. El objetivo de este trabajo es emplear teledetección y clasificación digital de imágenes para obtener cartografías de usos y coberturas del suelo en la Demarcación Hidrográfica del Segura, para el periodo comprendido entre 1986 y 2019; además, los resultados se analizarán para conocer los cambios más significativos que haya experimentado la superficie terrestre del área de estudio. Para ello se ha utilizado el algoritmo Random Forest con imágenes de Landsat 5, 7 y 8; aunque la exactitud global conseguida indica que existe un margen de mejora en el ajuste del modelo, las clasificaciones obtenidas se consideran fiables. Respecto a los usos y coberturas del suelo, se obtiene una disminución del uso forestal y un aumento de los usos agrícolas, la superficie cubierta por matorral, los suelos desnudos y los invernaderos.

Indo Gangetic flood plains are subjected to recurring floods due to large low-lying land areas adjacent to major Himalayan Rivers. Recent floods in the Ganga River during the late withdrawal phase of Indian summer monsoon rains in 2019 had reportedly caused death of ∼200–500 people and damaged major infrastructures in different parts of Bihar state, India. This paper presents a detailed analysis of the spatial extent and temporal pattern of flood inundation areas in Patna district of Bihar using Sentinel-1 Synthetic Aperture Radar (SAR) data. The results obtained using K-means classification show that about 1268 km2 (20% of the studied portion) of the land area were inundated during the peak flooding conditions on 9th October 2019. The peak flooding lasts for more than 20 days, with a maximum water depth of 18 m in the main river. Owing to the continually changing nature of river morphology by recurring floods, we also analyzed the channel morphology and bar characteristics before, during, and after the floods. Geomorphic change assessment indicates that the total bar occupied area was decreased from 661 km2 for pre-flood time to 446 km2 in peak flood time, whereas it retains its original shape in post-flood images.

The article presents the concept of a maintenance-free inspection station intended for conducting drive-thru tests of commercial vehicles. The main purpose of building this type of diagnostic line is to carry out non-invasive, preliminary tests of heavy-duty vehicles entering the vehicle service area in terms of parameters affecting the safety of their operation in relation to the applicable standards. The main parameter to be assessed will be the concentration of toxic exhaust components, measured using remote sensing methods. In addition, the proposed diagnostic line can be supplemented with additional remote measurement systems, such as, for example, systems for assessing the condition of vehicle lighting, loads on individual axles and individual wheels of the vehicle, tire pressure, thermal load of the brake system, as well as a system for detecting leaks of fluids from the vehicle. Based on the carried out work, it has been shown that using the current specialist knowledge and the components of measurement systems available on the market, it is possible to develop an innovative diagnostic line using remote measurement methods.

Vehicle detection in remote sensing images remains a challenge because most vehicles are small and cover only a relatively small area due to the low ground sample distance. Although image super-resolution can improve small object detection performance as a preprocessing step, methods for improving the quality of the entire image tend to focus on the majority backgrounds that are not important for detection and involve high computational cost. Inspired by the promising feature-level super-resolution method, in this letter, we propose a novel anchor-free vehicle detection network for small vehicle detection in remote sensing images. Specifically, a target-guided feature super-resolution network is proposed to enhance the features of the potential target. Besides, we propose a novel feature fusion module to improve the feature representation of shallow layers, which accounts for small object detection. Extensive experiments on three public remote sensing detection datasets [cars overhead with context (COWC), Vehicle Detection in Aerial Imagery (VEDAI), and UCAS-AOD] amply demonstrate that our method can achieve significant performance with a mean average precision of 0.933, 0.756, and 0.961, respectively.

Missing data reconstruction is a classical yet challenging problem in remote sensing images. Most current methods based on traditional convolutional neural network require supplementary data and can only handle one specific task. To address these limitations, we propose a novel generative adversarial network-based missing data reconstruction method in this letter, which is capable of various reconstruction tasks given only single source data as input. Two auxiliary patch-based discriminators are deployed to impose additional constraints on the local and global regions, respectively. In order to better fit the nature of remote sensing images, we introduce special convolutions and attention mechanism in a two-stage generator, thereby benefiting the tradeoff between accuracy and efficiency. Combining with perceptual and multiscale adversarial losses, the proposed model can produce coherent structure with better details. Qualitative and quantitative experiments demonstrate the uncompromising performance of the proposed model against multisource methods in generating visually plausible reconstruction results. Moreover, further exploration shows a promising way for the proposed model to utilize spatio-spectral-temporal information. The codes and models are available at https://github.com/Oliiveralien/Inpainting-on-RSI.

Copyright: 2020 IEEE. This is a pre-print version. The definitive version of the work is published in Mdakane, L.W. & Kleynhans, W. 2020. Feature selection and classification of oil spill from vessels using Sentinel-1 wide–swath synthetic aperture radar data. IEEE Geoscience and Remote Sensing Letters, pp. 12

The acquisition of road information from remote sensing images is of significant value with regard to intelligent transportation research. This study focuses on enhancing the contour-learning ability to mitigate the phenomenon of fragmented road segments and missing connections in road extraction. A novel Deep Feature-Review (FR) Transmit Network (TransNet) is proposed to review and facilitate the flow of contour features into an encoder network. Meanwhile, multiscale features are linked via a bridge between the encoder and the decoder. Compared with the state-of-the-art models such as fully convolutional network (FCN), SegNet, DeepLabv3, D-LinkNet, spatial consistency-FCN, and generative adversarial network (GAN), the proposed network achieves better overall performance for the Massachusetts Roads data set, with accuracy, precision, recall, and mean intersection-over-union (IoU) scores of 97.48%, 83.72%, 78.13%, and 0.6286%, respectively. For the DeepGlobe Road Extraction data set, the proposed network outperforms FCN, SegNet, DeepLabv3, D-LinkNet, and Deep TransNet, achieving accuracy, precision, recall, and mean IoU scores of 98.70%, 87.30%, 81.15%, and 0.7244%, respectively. Overall, these experiments indicate that the proposed network can effectively address the phenomenon of fragmented road segments and poor connectivity in remote sensing images, indicating its potential for utilization in practical intelligent transportation scenarios.

Limited by the GPU memory, the current mainstream detectors fail to directly apply to large-scale remote sensing images for object detection. Moreover, the scale range of objects in remote sensing images is much wider than that of general images, which also greatly hinders the existing methods to effectively detect geospatial objects of various scales. For achieving high-performance object detection on large-scale remote sensing images, this article proposes a much faster and more accurate detecting framework, called cropping region proposal network-based scale folding network (CRPN-SFNet). In our framework, the CRPN includes a weak semantic RPN for quickly locating interesting regions and a strategy of generating cropping regions to effectively filter out meaningless regions, which can greatly reduce the computation and storage burden. Meanwhile, the proposed SFNet leverages the scale folding-based training and testing methods to extend the valid detection range of existing detectors, which is beneficial for detecting remote sensing objects of various scales, including very small and very large geospatial objects. Extensive experiments on the public Dataset for Object deTection in Aerial images data set indicate that our CRPN can help our detector deal the larger image faster with the limited GPU memory; meanwhile, the SFNet is beneficial to achieve more accurate detection of geospatial objects with wide-scale range. For large-scale remote sensing images, the proposed detection framework outperforms the state-of-the-art object detection methods in terms of accuracy and speed.

Accurate object detection in remote sensing images is an essential part of automatic extraction, analysis, and understanding of image information, which potentially plays a significant role in a number of practical applications. However, the scale diversity in remote sensing images presents a substantial challenge for object detection, regarded as one of the crucial problems to be solved. To extract multiscale feature representations and sufficiently exploit semantic context information, this letter proposes a semantic context-aware network (SCANet) model for multiscale object detection. We propose two novel modules, called receptive field-enhancement module (RFEM) and semantic context fusion module (SCFM), to enhance the performance of SCANet. The RFEM dedicates to more robust multiscale feature extraction by paying attention to distinct receptive fields through multibranch different convolutions. For the purpose of utilizing the semantic context information contained in the scene to guide the network to better detection accuracy, the SCFM integrates the semantic context features from the upper level with the lower level features and delivers them hierarchically. Experiments demonstrate that, compared with the state-of-the-art approaches, the SCANet yields superior detection results on the DOTA-v1.5 data set.

The timely and accurate assessment of changes in mountain vegetation biomass and other parameters is of great importance to mountain ecosystem conservation. With the rapid development of remote sensing technology, hyperspectral remote sensing images have facilitated the large-scale and long-time series monitoring of environmental changes in mountainous areas. However, topographic effects cause remote sensing images of mountainous areas to be prone to spectral variations within the same land cover and spectral confusion among different land covers. This phenomenon seriously affects the accuracy of remote sensing inversions and hinders the development and application of remote sensing in mountainous areas. Numerous scholars have established various topographic correction models (TCMs) to eliminate the influence of topographic effects. Comparative evaluation of the performance of different TCMs allows us to better understand their characteristics. Most previous evaluation studies have directly applied in remote sensing images, which were limited by the changing conditions of the study area. Therefore, this letter used computer simulations to controllably evaluate six popular TCMs on hyperspectral images. The results showed that their performance varied with the spectral band, and overall, the best performance was achieved by the C correction model, followed by the sun-canopy-sensor (SCS) + C model. This letter provides a basis for the optimal selection of TCMs in complex terrains.

Few-shot image classification has attracted extensive attention, which aims to recognize unseen classes given only a few labeled samples. Due to the large intraclass variances and interclass similarity of remote sensing scenes, the task under such circumstance is much more challenging than general few-shot image classification. Most existing prototype-based few-shot algorithms usually calculate prototypes directly from support samples and ignore the validity of prototypes, which results in a decline in the accuracy of subsequent inferences based on prototypes. To tackle this problem, we propose a Siamese-prototype network (SPNet) with prototype self-calibration (SC) and intercalibration (IC). First, to acquire more accurate prototypes, we utilize the supervision information from support labels to calibrate the prototypes generated from support features. This process is called SC. Second, we propose to consider the confidence scores of the query samples as another type of prototypes, which are then used to predict the support samples in the same way. Thus, the information interaction between support and query samples is implicitly a further calibration for prototypes (so-called IC). Our model is optimized with three losses, of which two additional losses help the model to learn more representative prototypes and make more accurate predictions. With no additional parameters to be learned, our model is very lightweight and convenient to employ. The experiments on three public remote sensing image datasets demonstrate competitive performance compared with other advanced few-shot image classification approaches. The source code is available at https://github.com/zoraup/SPNet.

Remote sensing image super-resolution (SR) plays an important role by supplementing the lack of original high-resolution (HR) images in the study scenarios of large spatial areas or long time series. However, due to the lack of imagery information in low-resolution (LR) images, single-image super-resolution (SISR) is an inherently ill-posed problem. Especially, it is difficult to reconstruct the fine textures of HR images at large upscaling factors (e.g., four times). In this work, based on Google Earth HR images, we explore the potential of the reference-based super-resolution (RefSR) method on remote sensing images, utilizing rich texture information from HR reference (Ref) images to reconstruct the details in LR images. This method can use existing HR images to help reconstruct the LR images of long time series or a specific time. We build a reference-based remote sensing SR data set (RRSSRD). Furthermore, by adopting the generative adversarial network (GAN), we propose a novel end-to-end reference-based remote sensing GAN (RRSGAN) for SR. RRSGAN can extract the Ref features and align them to the LR features. Eventually, the texture information in the Ref features can be transferred to the reconstructed HR images. In contrast to the existing RefSR methods, we propose a gradient-assisted feature alignment method that adopts the deformable convolutions to align the Ref and LR features and a relevance attention module (RAM) to improve the robustness of the model in different scenarios (e.g., land cover changes and cloud coverage). The experimental results demonstrate that RRSGAN is robust and outperforms the state-of-the-art SISR and RefSR methods in both quantitative evaluation and visual results, which indicates the great potential of the RefSR method for remote sensing tasks. Our code and data are available at https://github.com/dongrunmin/RRSGAN.

Accurate objects detection in remote sensing images is very important, because security, transportation, and rescue applications in military and civilian fields require fully analyzing and using these images. To address the problem that many small-sized objects in remote sensing images are difficult to detect, this paper proposes an improved S 2 ANET-SR model based on S 2 A-NET network. In this paper, the original and reduced image are fed to the detection network at the same time, and then a super-resolution enhancement module for the reduced image is designed to enhance the feature extraction of small objects, after that, the perceptual loss and texture matching loss is proposed as supervision. Extensional experiments are conducted to evaluate the performance on the general remote sensing dataset DOTA, and the results show that our proposed method can achieve 74.47% mAP, which is 0.79% better than the accuracy of S 2 A-NET.

Scene classification in very high-resolution (VHR) remote sensing (RS) images is a challenging task due to the complex and diverse content of the images. Recently, convolution neural networks (CNNs) have been utilized to tackle this task. However, CNNs cannot fully meet the needs of scene classification due to clutters and small objects in VHR images. To handle these challenges, this letter presents a novel multilevel feature fusion (MLFF) network with adaptive channel dimensionality reduction for RS scene classification. Specifically, an adaptive method is designed for channel dimensionality reduction of high-dimensional features. Then, an MLFF module is introduced to fuse the features in an efficient way. Experiments on three widely used data sets show that our model outperforms several state-of-the-art methods in terms of both accuracy and stability.

Convolutional neural networks (CNNs) have achieved remarkable results in semantic segmentation of high-resolution remote sensing images (HRRSIs). However, the scales and textures of HRRSIs are diverse, which makes it difficult for a fixed-layer CNN to obtain rich features. In this regard, we propose an end-to-end ensemble fully convolutional network (EFCNet), which mainly includes two modules: the adaptive fusion module (AFM) and the separable convolutional module (SCM). The AFM can fuse features of different scales based on ensemble learning, whereas the SCM can reduce the complexity of the model under multifeature fusion. In the experiment, we use UNet and PSPNet to verify the framework on the ISPRS Vaihingen and Potsdam datasets. The experimental results show that the EFCNet can effectively improve the final segmentation performance and reduce the complexity of the ensemble model.

With the introduction of artificial intelligence (AI), current deep learning-driven change detection methods generally regard changes as a type of specific land cover information and try to detect them simply by using existing semantic labeling models. In practice, the changes that occur on different land cover categories may appear completely different for remote sensing images, making it difficult to detect multiple categories of changes with an end-to-end model. In addition, training such networks requires a large amount of prelabeled references, which are labor-intensive and time-consuming. Motivated by this observation, we integrate dual learning algorithm and disentangled representation theory to develop a novel approach, named content-invariant dual learning (CiDL), for either supervised or unsupervised change detection in remote sensing images. In our framework, two opposite Y-shaped networks, each of which consists of two encoders and one decoder, are introduced to translate bitemporal images from their original domains to each others, where their intrinsic content features are retained, while their style features are consistent. By training our hybrid framework, even without references, this method learns a category-wise cross-domain translation to suppress the discrepancies in paired unchanged regions, and meanwhile highlight those in paired changed regions. The experimental results on two typical change detection data sets and the comparison with other state-of-the-art deep learning-driven methods verify the effectiveness and competitiveness of our proposed CiDL.