Search

Both local and global context dependencies are essential for building extraction from remote sensing (RS) images. Convolutional neural network (CNN) can extract local spatial details well but lacks the ability to model long-range dependency. In recent years, vision transformer (ViT) has shown great potential in modeling global context dependency. However, it usually brings huge computational cost, and spatial details cannot be fully retained in the process of feature extraction. To maximize the advantages of CNNs and ViTs, we propose dual spatial attention transformer net (DSAT-Net), which combines them in one model. In DSAT-Net, we design an efficient dual spatial attention transformer (DSAFormer) to solve the defects of standard ViT. It has a dual attention structure to complement each other. Specifically, the global attention path (GAP) conducts large-scale downsampling of the feature maps before the global self-attention (SA) computing, to reduce the computational cost. The local attention path (LAP) uses efficient stripe convolution to generate local attention, which can alleviate the loss of information caused by downsampling operation in the GAP and supplement the spatial details. In addition, we design a feature refining module called channel mixing feature refine module (CM-FRM) to fuse low- and high-level features. Our model achieved competitive results on three public building extraction datasets. The code will be available at https://github.com/stdcoutzrh/BuildingExtraction. [ABSTRACT FROM AUTHOR]

In this letter, we introduce deep active learning (AL) for multi-label image classification (MLC) problems in remote sensing (RS). In particular, we investigate the effectiveness of several AL query functions for the MLC of RS images. Unlike the existing AL query functions (which are defined for single-label classification or semantic segmentation problems), each query function in this letter is based on the evaluation of two criteria: 1) multi-label uncertainty and 2) multi-label diversity. The multi-label uncertainty criterion is associated with the confidence of the deep neural networks (DNNs) in correctly assigning multi-labels to each image. To assess this criterion, we investigate three strategies: 1) learning multi-label loss ordering; 2) measuring temporal discrepancy of multi-label predictions; and 3) measuring the magnitude of approximated gradient embeddings. The multi-label diversity criterion is associated with the selection of a set of images that are as diverse as possible to each other which prevents redundancy among them. To assess this criterion, we exploit a clustering-based strategy. We combine each of the abovementioned uncertainty strategies with the clustering-based diversity strategy, resulting in three different query functions. All the considered query functions are introduced for the first time in the framework of MLC problems in RS. Experimental results obtained on two benchmark archives show that these query functions result in the selection of a highly informative set of samples at each iteration of the AL process. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have been widely used in the field of remote sensing (RS) scene classification, which have achieved remarkable results. In RS scene classification, local key objects are particularly crucial for classification results. However, most existing CNN methods directly utilize the deep-level global features of CNN, ignoring object-level information in shallow features or leading to redundant and erroneous information when using shallow features. To fully utilize the important information in shallow features, we proposed an end-to-end contextual spatial-channel attention network (CSCANet) to learn multilayer feature representations and further improve classification performance by employing shallow object-level semantic information. First, ResNet34 is pretrained to extract different levels of features. Second, a contextual spatial-channel attention module (CSCAM) is constructed to generate contextual spatial-channel attention features by exploiting features at different levels. Finally, the triple loss function is combined with the central loss function to guide the model training. Experiments on three public RS scene classification datasets [UC-Merced (UCM), aerial image dataset (AID), and NWPU-RESISC45 (NWPU)] demonstrate that the proposed method achieves highly competitive results. [ABSTRACT FROM AUTHOR]

Forest inventory forms the foundation of forest management. Remote sensing (RS) is an efficient means of measuring forest parameters at scale. Remotely sensed species classification can be used to estimate species abundances, distributions, and to better approximate metrics such as aboveground biomass. State-of-the-art methods of RS species classification rely on deep-learning models such as convolutional neural networks (CNNs). These models have two major drawbacks: they require large samples of each species to classify well and they lack explainability. Therefore, rare species are poorly classified causing poor approximations of their associated parameters. We show that the classification of rare species can be improved by as much as eight F1-points using a neuro-symbolic (NS) approach that combines CNNs with an NS framework. The framework allows for the incorporation of domain knowledge into the model through the use of mathematically represented rules, improving model explainability. [ABSTRACT FROM AUTHOR]

Benefiting from effective global information interaction, vision transformers (ViTs) have been widely used in the building extraction task. However, buildings in remote sensing (RS) images usually differ greatly in size. Mainstream ViT-based segmentation models for RS images are based on Swin Transformer, which lacks multiscale information inside the ViT block. In addition, they only connect the output of the entire ViT encoder block to the decoder, which ignore the similarity information of the attention maps inside the ViT encoder block and are unable to provide better global dependencies for the decoder. To solve the above problems, we introduce a novel shunted transformer, which enables the model to capture multiscale information internally while fully establishing global dependencies, to build a pure ViT-based U-shaped model for building extraction. Furthermore, unlike the previous single-skip-connection structure of the U-shaped methods, we build a novel dual skip connection structure inside the model. It simultaneously transmits the attention maps inside the ViT encoder block and its entire output to the decoder, thereby fully mining the information of the ViT encoder block and providing better global information guidance for the decoder. Thus, our model is named shunted dual skip connection UNet (SDSC-UNet). We also design a feature fusion module called dual skip upsample fusion module (DSUFM) to aggregate the information. Our model yields the state-of-the-art (SOTA) performance [83.02% intersection over union (IoU)] on the Inria Aerial Image Labeling Dataset. Code will be available at https://github.com/stdcoutzrh/BuildingExtraction. [ABSTRACT FROM AUTHOR]

Recently, convolutional neural networks (CNNs) have been developed for remote sensing image fusion (RSIF). To obtain competitive fusion performance, network design becomes more complicated by stacking convolutional layers deeper and wider. However, problems still remain when applying the existing networks in practical applications. On the one hand, researchers focus on improving spatial resolution but ignore that the fused images will be used in subsequent interpretation applications, e.g., objection detection. On the other hand, RSIF involves different tasks with different image sources, e.g., pansharpening of the panchromatic and multispectral image (MSI), hypersharpening of the panchromatic and hyperspectral image (HSI), and so on. However, the existing networks only solve one of them, failing to be compatible with other tasks. To address the above problems, a convenient task-inspired multiscale nonlocal-attention network (MNAN) is proposed for RSIF. The proposed MNAN focuses more on enhancing the multiscale targets in the scene when improving the resolution of the fused image. In addition, the proposed network can be applied to both pansharpening and hypersharpening tasks without any modification. [ABSTRACT FROM AUTHOR]

The diversity and complexity of the subsurface, as well as the similarity of cloud features to those of highlighted surface objects (especially snowy areas), are the main reasons for the current limitations in cloud detection accuracy. This letter proposes a semantic segmentation network fusing geographic features with contextual information for cloud detection named GCI_CD. We design three modules: geographical-cloud information fusion module (GCIFM) (to distinguish between clouds and snow), separable ResNet with CSAM encoder (to enhance channel and spatial association information extraction), and contextual information integration module (CIIM) (to extract multiscale features). To the best of our knowledge, this is the first time that geographic information has been fused in a cloud detection method. The validity of our proposed method is demonstrated using a high-resolution GF-1 dataset containing the near-infrared band and dividing the snow-covered and non-snow-covered areas. The experimental results show that GCI_CD has a better performance not only in snow-covered regions but also in the non-snow-covered regions. Especially for snow-covered images, the IOU of GCI_CD’s cloud detection results are improved by 5% on average compared to other cloud detection methods. [ABSTRACT FROM AUTHOR]

Extraction road from remote sensing (RS) images is a challenging topic because of the inhomogeneous intensity, nonconsistent contrast, and very cluttered background of satellite images. Most previous approaches, relying on convolutions or self-attention, are built on the local operation or global modeling on the spatial domain but are difficult to capture weak and continuous road objects. The spectral representation of road image features and modulation learning on it provides a novel long-range-dependent and fine-grained feature representation mechanism. Based on it, we propose a novel road extraction network on RS images, called an adaptive Fourier filtered U-shaped network (AFU-Net) in this letter, which relies on modulation learning on the Fourier domain. The AFU-Net is composed of modulation learner (MoL) basic blocks and follows the pipeline of the classical U-Net model. The basic MoL block includes a global MoL (GML) block for global spectral modulation learning and an attentive MoL (AML) block which contains two parallel layers, i.e., phase-modulated filter (PMF) and magnitude-modulated filter (MMF), for fine-grained spectral modulation on the Fourier spectrum. The experiments on two public datasets, such as Massachusetts roads and DeepGlobe road datasets have shown the outstanding performance of AFU-Net on the metrics of accuracy, precision, recall, and mean intersection over union (mIoU). [ABSTRACT FROM AUTHOR]

Old urban reform usually reactivates the urban fabric in a new era of sustainable development. However, what remains of the former fabric and how it interacts with the new one often inspires curiosity. How the old residents adapt their lives to the new layout should be explored qualitatively and quantitatively. This research aimed to assess the old and new fabrics in the downtown area of Keelung, Taiwan, by considering the interactions between truncated layout, proportion, and infill orientation in the mature and immature interfaces. According to the historical reform map made in 1907, the newly constructed area occupied the old constructed area in seven downtown blocks. On average, the area composed of new buildings ranged from 135.60% to 239.20% of the old area, and the average volume of the buildings reached a maximum of 41.72 m when compared to the old buildings in place prior to the reform. It seems that the new fabric purposefully maintained the old temples at the centers of the blocks. However, the old alleys, which still remain within these blocks, have been significantly overloaded with services and have become auxiliary utility spaces for the in-block residences. With regard to the part of the fabric that was truncated or reoriented by new streets, the modification could also be easily found on the second skin. A physical model analysis used a UAV 3D cloud model and QGIS® to verify the axes, hierarchies, entrances, open spaces, and corners in the commission store block and temple blocks. We found that the 3D point model and historical maps presented a convincing explanation of the evolved fabric from the past to the present. The stepwise segmentation visualizes the enclosed block inside a block on the historical maps and according to the present sections. We found that new roles for old alleys have evolved behind the new fabric. [ABSTRACT FROM AUTHOR]

Deep learning plays a highly essential role in the domain of remote sensing change detection (CD) due to its high efficiency. From some existing methods, we can observe that the fusion of information at each scale is quite vital for the accuracy of the CD results, especially for the common problems of pseudo-change and the difficult detection of change edges in the CD task. With this in mind, we propose a New Fusion network with Dual-branch Encoder and Triple-branch Decoder (DETDNet) that follows a codec structure as a whole, where the encoder adopts a siamese Res2Net-50 structure to extract the local features of the bitemporal images. As for the decoder in previous works, they usually employed a single branch, and this approach only preserved the fusion features of the encoder's bitemporal images. Distinguished from these approaches, we adopt the triple-branch architecture in the decoder for the first time. The triple-branch structure preserves not only the dual-branch features from the encoder in the left and right branches, respectively, to learn the effective and powerful individual features of each temporal image but also the fusion features from the encoder in the middle branch. The middle branch utilizes triple-branch aggregation (TA) to realize the feature interaction of the three branches in the decoder, which enhances the integrated features and provides abundant and supplementary bitemporal feature information to improve the CD performance. The triple-branch architecture of the decoder ensures that the respective features of the bitemporal images as well as their fused features are preserved, making the feature extraction more integrated. In addition, the three branches employ a multiscale feature extraction module (MFE) per layer to extract multiscale contextual information and enhance the feature representation capability of the CD. We conducted comparison experiments on the BCDD, LEVIR-CD, and SYSU-CD datasets, which were created in New Zealand, the USA, and Hong Kong, respectively. The data were preprocessed to contain 7434, 10,192, and 20,000 image pairs, respectively. The experimental results show that DETDNet achieves F1 scores of 92.7%, 90.99%, and 81.13%, respectively, which shows better results compared to some recent works, which means that the model is more robust. In addition, the lower FP and FN indicate lower error and misdetection rates. Moreover, from the analysis of the experimental results, compared with some existing methods, the problem of pseudo-changes and the difficulty of detecting small change areas is better solved. [ABSTRACT FROM AUTHOR]

Remote sensing image change detection (RS-CD) has made impressive progress with the help of deep learning techniques. Small object change detection (SoCD) still faces many challenges. On the one hand, when the scale of changing objects varies greatly, deep learning models with overall accuracy as the optimization goal tend to focus on large object changes and ignore small object changes to some extent. On the other hand, the RS-CD model based on deep convolutional networks needs to perform multiple spatial pooling operations on the feature map to obtain deep semantic features, which leads to the loss of small object feature-level information in the local space. Therefore, we propose a Siamese transformer change detection network with a multiscale window via an adaptive fusion strategy (SWaF-Trans). To solve the problem of ignoring small object changes, we compute self-attention in windows of different scales to model changing objects at the corresponding scales and establish semantic information links through a moving window mechanism to capture more comprehensive small object features in small-scale windows, thereby enhancing the feature representation of multiscale objects. To fuse multiscale features and alleviate the problem of small object feature information loss, we propose a channel-related fusion mechanism to model the global correlation between channels for display and adaptively adjust the fusion weights of channels to enable the network to capture more discriminative features of interest and reduce small object feature information loss. Experiments on the CDD and WHU-CD datasets show that SWaF-Trans exceeds eight advanced baseline methods, with absolute F1 scores as high as 97.10% and 93.90%, achieving maximum increases of 2% and 5.6%, respectively, compared to the baseline methods. [ABSTRACT FROM AUTHOR]

Remote sensing (RS) and geographic information systems (GIS) have been used to guide the selection of suitable areas for aquaculture. This systematic review synthesizes the key suitability factors and constraints reported in the literature for establishing inland pond aquaculture. We searched for primary studies on Scopus and Web of Science according to preferred reporting items for systemic reviews and meta-analysis (PRISMA) guidelines. Between 1991 and 2020, 354 articles were published in 104 academic journals. The maximum annual number of publications occurred in 2020, with 22 publications, and there is an increasing trend in studies published over the past 30 years. From 12 selected studies, we identified 48 suitability factors, 11 related to soil suitability, 19 to socioeconomic and infrastructure suitability, and 18 to water quality and availability. The most frequently used suitability factors were road proximity, local market center distance, soil texture, soil slope, and water temperature. We listed 15 constraints that restrict or limit the selection of specific geographic locations for inland aquaculture. Urbanized areas, roads, and forests were the most frequently restricted areas. Geotechnologies provide powerful tools for spatial planning and management of aquaculture. Availability, quality, and access to spatial data are critical for the use of geotechnologies in the process of aquaculture site selection. [ABSTRACT FROM AUTHOR]

[Display omitted] • GSREI regarding accumulated subsidence, subsidence rate, and elevation. • VEI from population, GDP, construction land, viaduct and metro density. • Risk assessment model and the GSHI index. Urban ground subsidence (GS) causes ground elevation loss, threatens the safe operation of various facilities, and affects surface runoff and hydrological circulation. Therefore, it is vital to monitor the current status of GS to predict and evaluate potential risks to sustainable urban development. Sentinel SAR images from 2019 to 2020 combined with the data from the 2019–2020 Statistical Yearbook were used as data sources. Based on PS-InSAR and SBAS-InSAR, GIS was integrated with the AHP method. A ground subsidence risk evaluation index (REI) suitable for Shanghai was developed regarding three aspects of the amount of accumulated land subsidence (RS obtained): subsidence rate, and terrain elevation. The vulnerability evaluation index (VEI) was established from five aspects: population density, GDP per unit area, the proportion of construction land, viaduct density, and metro density. We introduced the fuzzy AHP method to complete the risk assessment of ground subsidence in Shanghai and obtained the ground subsidence hazard index (GSHI). We used GIS superimposed remote sensing images to visualize the individual evaluation indicators REI , VEI , and GSHI. Using GIS superimposed remote sensing images, various evaluation indicators and GSHI were visualized. Based on the evaluation results, targeted prevention and control recommendations are provided. The results of this research can provide a basis for scientific decision-making regarding Shanghai's territorial spatial planning and socio-economic development. [ABSTRACT FROM AUTHOR]

Automatic and accurate extraction of land use/cover change (LUCC) is essential for various applications, particularly studies on climate change and sustainable development. However, the automatic and accurate detection of LUCC in large-scale regions remains challenging due to the complexity of LUCC and the high cost of training data acquisition. To address this research gap, a simple and practical unsupervised change detection method based on multi-indices and bitemporal remote sensing image pairs (USCD-MiBi) was proposed to automatically extract LUCC in two heterogeneous experimental sites without training data. Single-index change analysis (SICA), multi-index change analysis (MICA), and bitemporal change analysis (BTCA) form the core of the USCD-MiBi method. In this method, all the selected change indices were integrated by the SICA and MICA to identify potential change regions. The potential false changes, caused by inconsistencies in atmospheric conditions and phenology, were further removed by the BTCA. Verification experiments revealed that the detection accuracy of USCD-MiBi method can exceed 90%, whether the LUCC is concentrated in urban regions or scattered in mountainous areas. It was also observed that the proposed method had higher detection performance than several other change detection methods. The proposed USCD-MiBi method offers high flexibility and adaptability, allowing users to choose the most suitable change indices based on the characteristics of their study area or the selected data sources. This study provides a simple solution for automatically detecting large-scale LUCC without training samples, making it accessible to a wider range of users.

Transformer has shown excellent performance in remote sensing field with long-range modeling capabilities. Remote sensing video (RSV) moving object detection and tracking play indispensable roles in military activities as well as urban monitoring. However, transformers in these fields are still at the exploratory stage. In this survey, we comprehensively summarize the research prospects of transformers in RSV moving object detection and tracking. The core designs of remote sensing transformers and advanced transformers are first analyzed. It mainly includes the attention mechanism evolution for specific tasks, the fitting ability design of input mapping, diverse feature representation, model optimization, etc. The architectural characteristics of RSV detection and tracking are then described across two aspects. One is moving object detection for motion-based traditional background subtractions and appearance-based deep learning models. The other is object tracking for single and multiple targets. The research difficulties mainly include the blurred foreground in RSV data, the irregular object movement in traditional background subtraction, and the severe object occlusion in object tracking. Following that, the potential significance of transformers is discussed according to some thorny problems in RSV. Finally, we summarize ten open challenges of transformers in RSV, which may be used as a reference for promoting future research.

Change detection (CD) is a major topic in remote sensing research. Deep learning (DL)-based CD methods have made great progress. However, existing CD methods have difficulty in exploiting the different semantic and detailed information in deep and shallow features, which often leads to blurred target boundaries of identified changes. In addition, most CD methods based on the NestedUNet structure focus on improving accuracy, ignoring the importance of efficiency. Therefore, in this article, an adaptive fusion NestedUNet for CD (AFNUNet) is proposed. AFNUNet compresses the model parameters and computational cost by using the encoder based on the inverted bottleneck structure and the decoder based on the depthwise convolution. The correlation between the final multilevel feature maps extracted by NestedUNet is difficult to model by summation or concatenation. Therefore, an attention mechanism-based adaptive fusion module (AFM) is proposed. The AFM allows the network to adaptively select feature information from the final different layers of features extracted from NestedUNet in both channel and spatial dimensions so that the fused features capture deep rich semantic information while retaining detailed information at shallow boundaries. Finally, a loss function based on the Bray-Curtis distance is introduced for suppressing the sample imbalance problem. Extensive experiments on the WHU-CD, LEVIR-CD, and SYSU-CD datasets demonstrate that AFNUNet surpasses several state-of-the-art (SOTA) CD methods in terms of effectiveness. Moreover, the proposed AFNUNet remarkably reduces Params and FLOPs by 63% and 70% compared to other NestedUNet-based CD models.

As an atmospheric phenomenon, haze significantly reduces the visibility of outdoor and remote sensing images. As remote sensing and outdoor imaging have different mechanisms, existing dehazing methods are hard to be applied for both outdoor images and remote sensing images. In this article, an efficient dehazing method is proposed which can be applied to both outdoor and remote sensing images. The proposed method has the advantages of both dehazing methods based on image enhancement and image restoration methods. To address the problem of inaccurate calculation of transmittance in existing methods, a Gaussian-weighted image fusion is introduced to obtain a refined transmittance. In addition, to solve the problem of image color distortion after dehazing, an unsharp mask method is used to correct the dehazed image. Experiments on the synthetic dataset and real dataset show that the proposed method is able to dehaze both the outdoor and remote sensing images and outperforms existing methods by visual inspection. On the RICE remote sensing image dataset, the proposed method has an image peak signal-to-noise ratio of 27.08 and structural similarity of 0.94, which are higher than other methods.

Visual question generation (VQG) is a fundamental task in vision-language understanding that aims to generate relevant questions about the given input image. In this article, we propose a paragraph-based VQG approach for generating intelligent questions in natural language about remote sensing (RS) images. Specifically, our proposed framework consists of two transformer-based vision and language models. First, we employ a swin-transformer encoder to generate a multiscale representative visual feature from the image. Then, this feature is used as a prefix to guide a generative pretrained transformer-2 (GPT-2) decoder in generating multiple questions in the form of a paragraph to cover the abundant visual information contained in the RS scene. To train the model, the language decoder is fine-tuned on RS dataset to generate a set of relevant questions from the RS image. We evaluate our model on two visual question-answering (VQA) datasets in RS. In addition, we construct a new dataset termed TextRS-VQA for better evaluation for our VQG model. This dataset consists of questions completely annotated by humans which addresses the high redundancy of the questions in prior VQA datasets. Extensive experiments using several accuracy and diversity metrics demonstrate the effectiveness of our proposed VQG model in generating meaningful, valid, and diverse questions from RS images.

Sparse representation-based single image super-resolution (SISR) methods use a coupled overcomplete dictionary trained from high-resolution images/image patches. Since remote sensing (RS) satellites capture images of large areas, these images usually have poor spatial resolution and obtaining an effective dictionary as such would be very challenging. Moreover, traditional patch-based sparse representation models for reconstruction tend to give unstable sparse solution and produce visual artefact in the recovered images. To mitigate these problems, in this article, we have proposed an adaptive joint sparse representation-based SISR method that is dependent only on the input low-resolution image for dictionary training and sparse reconstruction. The new model combines patch-based local sparsity and group sparse representation-based nonlocal sparsity in a single framework, which helps in stabilizing the sparse solution and improve the SISR results. The experimental results are evaluated both visually and quantitatively for several RGB and multispectral RS datasets, where the proposed method shows improvements in peak signal-to-noise ratio by 1–4 dB and 2–3 dB over the state-of-the-art sparse representation- and deep learning-based SR methods, respectively. Land cover classification applied on the super-resolved images further validate the advantages of the proposed method. Finally, for practical RS applications, we have performed parallel implementation in general purpose graphics processing units and achieved significant speed ups (30–40×) in the execution time.

Multisensor data analysis allows exploiting heterogeneous data regularly acquired by the many available remote sensing (RS) systems. Machine- and deep-learning methods use the information of heterogeneous sources to improve the results obtained by using single-source data. However, the state-of-the-art methods analyze either the multiscale information of multisensor multiresolution images or the time component of image time series. We propose a supervised deep-learning classification method that jointly performs a multiscale and multitemporal analysis of RS multitemporal images acquired by different sensors. The proposed method processes very-high-resolution (VHR) images using a residual network with a wide receptive field that handles geometrical details and multitemporal high-resolution (HR) image using a 3-D convolutional neural network that analyzes both the spatial and temporal information. The multiscale and multitemporal features are processed together in a decoder to retrieve a land-cover map. We tested the proposed method on two multisensor and multitemporal datasets. One is composed of VHR orthophotos and Sentinel-2 multitemporal images for pasture classification, and another is composed of VHR orthophotos and Sentinel-1 multitemporal images. Results proved the effectiveness of the proposed classification method.

Due to the wide application of remote sensing (RS) image scene classification, more and more scholars activate great attention to it. With the development of the convolutional neural network (CNN), the CNN-based methods of the RS image scene classification have made impressive progress. In the existing works, most of the architectures just considered the global information of the RS images. However, the global information contains a large number of redundant areas that diminish the classification performance and ignore the local information that reflects more fine spatial details of local objects. Furthermore, most CNN-based methods assign the same weights to each feature vector causing the mode to fail to discriminate the crucial features. In this article, a novel method by Two-branch Deep Feature Embedding (TDFE) with a dual attention-aware (DAA) module for RS image scene classification is proposed. In order to mine more complementary information, we extract global semantic-based features of high level and local object-based features of low level by the TDFE module. Then, to focus selectively on the key global-semantics feature maps as well as the key local regions, we propose a DAA module to attain those key information. We conduct extensive experiments to verify the superiority of our proposed method, and the experimental results obtained on two widely used RS scene classification benchmarks demonstrate the effectiveness of the proposed method.

Deep learning has become one of remote sensing scientists' most efficient computer vision tools in recent years. However, the lack of training labels for the remote sensing datasets means that scientists need to solve the domain adaptation (DA) problem to narrow the discrepancy between satellite image datasets. As a result, image segmentation models that are then trained, could better generalize and use an existing set of labels instead of requiring new ones. This work proposes an unsupervised DA model that preserves semantic consistency and per-pixel quality for the images during the style-transferring phase. This article's major contribution is proposing the improved architecture of the SemI2I model, which significantly boosts the proposed model's performance and makes it competitive with the state-of-the-art CyCADA model. A second contribution is testing the CyCADA model on the remote sensing multiband datasets, such as WorldView-2 and SPOT-6. The proposed model preserves semantic consistency and per-pixel quality for the images during the style-transferring phase. Thus, the semantic segmentation model, trained on the adapted images, shows substantial performance gain compared to the SemI2I model and reaches similar results as the state-of-the-art CyCADA model. The future development of the proposed method could include ecological domain transfer, a priori evaluation of dataset quality in terms of data distribution, or exploration of the inner architecture of the DA model.

ABSTRACTA precise map of the dispersion of the groundwater potential across each watershed can help decision-makers to exert optimal water management in each region. In this research, the potential of groundwater resources in both the Zanjanrood Catchment and the Tarom Region, located in the northwest of Iran, has been studied. Seven effective criteria including slope, land-use, drainage density, spring density, lithology, lineament density, and rainfall are considered. Criteria were first weighted using the Analytical Hierarchical Process (AHP) method and then overlaid by the Technique for Order Preferences by Similarity to Ideal (TOPSIS) model. Finally, the spatial zoning map of groundwater potential was obtained in four categories. A sensitivity analysis was performed to determine the influence of each criterion on the obtained map. The model was verified using both the spatial distribution of the high-discharged production wells and the geophysical-based geoelectric field surveys. The results indicate that the high-discharged wells (>40 l/s) in both regions are dispersed predominantly in the very good zone and, in several cases, in the good zone. Besides, the results from the two-dimensional models of resistivity and induced polarization of geoelectrical field survey are inappropriate agreement with those from the TOPSIS method. Notably, there is no suitable potential zone of groundwater in the surrounding highlands to be used in the future for drinking purposes since the highlands water supply is a strategic supply for drinking. The strategies employed in this study, the results of GIS modeling, and the geoelectrical analysis can be considered for sustainable development of similar arid and semi-arid areas since groundwater is considered as the main supplier of water in such regions.

Forest plays an important role in the climate of a country and ecosystem's balance. During the last decade, District Abbottabad of Pakistan has gone through extensive land-use changes due to accelerated development, urbanization, and agriculture. This study examines the change in forest cover due to environmental factors with the help of satellite images (Landsat, Sentinel) and their classification through supervised classification, as well as the application of change detection technique after classification using geographical information system (GIS) over the last three decades (1986–2019). The land-use transition matrix was calculated from the year 1986 to 2019. The result shows an overall significant increase of 3.17, 17.24, and 7.24% in the forest, vegetation, and built-up areas, respectively; whereas, water bodies and others (barren land) have decreased significantly by 0.69 and 26.96%, respectively. This study also provides an assessment of the incline of carbon stock from the forest area over the period of the study by 17.19%. HIGHLIGHTS An increase of 3.17, 17.24, and 7.24% in the forest, vegetation, and build-up areas, respectively.; Water bodies and others (barren land) have decreased significantly by 0.69 and 26.96%, respectively.; Incline of carbon stock from the forest area over the period of the study by 17.19%.;

The present research aims to find the groundwater prospective areas by using Geographic Information System and Remote Sensing based frequency ratio technique. The data sets that affect groundwater such as geomorphology, lithology, slope, soil, rainfall, drainage density, lineament density and land use/ land cover were taken under consideration for finding the groundwater prospective areas in the Bundelkhand region, India. Thematic maps of all the above-mentioned factors have been prepared from the remotely sensed data, data taken from the different organizations and departments. The weights of each attribute class of each thematic layer are assigned according to the frequency ratio technique. A total, 799 numbers of spring locations and discharge rate data of 108 numbers of well locations have been taken under consideration for obtaining the frequency ratio of each attribute class of every parameter used and testing the obtained result, respectively. Finally, all layers were combined by raster calculator command in ArcGIS software to develop the map of groundwater prospective zones in the Bundelkhand region. The produced groundwater prospective areas were classified into five different zones such as very high, high, moderate, low and very low prospective zones and verified by actual well yield data.

In this paper, we present a lightweight and effective change detection model, called TinyCD. This model has been designed to be faster and smaller than current state-of-the-art change detection models due to industrial needs. Despite being from 13 to 140 times smaller than the compared change detection models, and exposing at least a third of the computational complexity, our model outperforms the current state-of-the-art models by at least 1 % on both F1-score and IoU on the LEVIR-CD dataset, and more than 8 % on the WHU-CD dataset. To reach these results, TinyCD uses a Siamese U-Net architecture exploiting low-level features in a globally temporal and locally spatial way. In addition, it adopts a new strategy to mix features in the space-time domain both to merge the embeddings obtained from the Siamese backbones, and, coupled with an MLP block, it forms a novel space-semantic attention mechanism, the Mix and Attention Mask Block (MAMB). [ABSTRACT FROM AUTHOR]

Mining activities degrade the landscape and ecosystems by introducing new land uses that alter soil characteristics. Mapping of this degradation is critical, particularly in the context of environmental protection, including climate change research. Even though mining provides significant industrial and economic benefits to society, it also decreases soil organic carbon (SOC) stocks and increases atmospheric carbon levels. This study aims to develop a practical method for determining the changes in SOC in the Altintepe Gold Mine, Ordu, Turkey, due to land cover changes caused by the mining process. 2013 and 2021 Landsat 8 images and CORINE data were used to map the land cover of the study area. NDVI analyses were conducted to detect land cover and SOC stocks, while the NDWI method was used to identify the water surfaces. SOC stock changes on the changed lands were calculated compared to the Turkey Soil Organic Carbon Stock Project database. The results showed that 109.85 ha of forest and 5.30 ha of agricultural land, corresponding to 4450.82 tons of SOC loss, were destroyed in the research area. Since the alterations in SOC levels are commonly determined by verifying remote sensing based analysis results with the site surveys, it becomes quite challenging to conduct such research in areas like Altintepe due to site access restrictions or data unavailability. From this point, this study presents a practical and alternative approach that avoids the necessity of fieldwork and provides a quick SOC change estimation based on the comparison of available data. The results are expected to provide a comprehensive and holistic perspective for the future operations and management of the mine sites and the surrounding environments. [ABSTRACT FROM AUTHOR]

One of the most common types of natural disaster, floods can happen anywhere on Earth, except in the polar regions. The severity of the damage caused by flooding can be reduced by putting proper management and protocols into place. Using remote sensing and a geospatial methodology, this study attempts to identify flood-vulnerable areas of the central district of Duhok, Iraq. The analytical hierarchy process (AHP) technique was used to give relative weights to 12 contributing parameters, including elevation, slope, distance from the river, rainfall, land use land cover, soil, lithology, topographic roughness index, topographic wetness index, aspect, the sediment transport index, and the stream power index in order to calculate the Flood Hazard Index (FHI). The relative importance of each criterion was revealed by a sensitivity analysis of the parameter values. This research developed a final flood susceptibility map and identified high-susceptible zones. This was classified anywhere from very low to very high classifications for its potential flood hazard. The generated map indicates that 44.72 km2 of the total land area of the study area in Duhok city has a very high susceptibility to flooding, and that these areas require significant attention from government authorities in order to reduce flood vulnerability. [ABSTRACT FROM AUTHOR]

This research aimed to know the tectonic activity of the Wadi Al-Batin alluvial fan using hydrological and morphotectonic analyses. Wadi Al-Batin alluvial fan is deposited from Wadi Al-Rimah in Saudi Arabia, which extended to Iraqi and Kuwait international boundaries. The longitudinal and transverse faults that characterize this region were common. The Abu- Jir-Euphrates faults have a significant impact on the region. The faults zone consists of several NW- SE trending faults running from the Rutba in western Iraq to the south along the Euphrates through Kuwait and meeting the Al-Batin fault to the Jal Al-Zor fault. The Hydromorphometric analysis of the present fan shows five watersheds having asymmetry shapes, more elongated and activity from the elongation ratio and asymmetry factor values. In contrast, transverse topographic symmetry and mountain front sinuosity factor have moderate to low activity from the stream length gradient index. The morphotectonic analysis suggests that the Wadi Al-Batin alluvial fan has a moderate tectonic activity due to the tectonic activity of the subsurface lineaments and the oil field activities found in the region. [ABSTRACT FROM AUTHOR]

We estimated the population density and quantified its characteristics using remote sensing, census data, and Geographic Information System (GIS). The interactive influence of these factors on population density was quantified based on geographic detectors to identify the differentiation mechanisms in the Chengdu metropolitan area of China. We identified the key factors that contribute to population density growth. The models used to simulate population density had the highest R2 values (>0.899). Population density tended to increase with time, with a multicentre spatial agglomeration pattern; the centre of gravity of the spatial distribution tended to move from the southeast to the northwest. Industry proportions, Normalised Difference Vegetation Index (NDVI), land use, distance to urban centers or construction land, and GDP per capita can satisfactorily explain population density changes. The combined impact of these elements on population density variation exhibited mutual and non-linear strengthening, with the mutual effect of the two elements intensifying the impact of each individual element. Our study identified the key driving forces that contribute to the differentiation of population density, which can provide valuable support for the development of effective regional and targeted population planning guidelines.

Landslide detection is crucial for disaster management and prevention. With the advent of multi-channel optical remote sensing technology, detecting landslides have become more accessible and more accurate. Although the use of the convolutional neural network (CNN) has significantly increased the accuracy of landslide detection on multi-channel optical remote sensing images, most previous methods using CNN lack the ability to obtain global context information due to the structural limitations of the convolution operation. Motivated by the powerful global modeling capability of the Swin transformer, we propose a new Conv-Trans Dual Network (CTDNet) based on Swin-Unet. First, we propose a dual-stream module (CTDBlock) that combines the advantages of ConvNeXt and Swin transformer, which can establish pixel-level connections and global dependencies from the CNN hierarchy to enhance the ability of the model to extract spatial information. Second, we apply an additional gating module (AGM) to effectively fuse the low-level information extracted by the shallow network and the high-level information extracted by the deep network and minimize the loss of detailed information when propagating. In addition, We conducted extensive subjective and objective comparison and ablation experiments on the Landslide4Sense dataset. Experimental results demonstrate that our proposed CTDNet outperforms other models currently applied in our experiments.

Contrastive learning allows us to learn general features for downstream tasks without the need for labeled data by leveraging intrinsic signals within remote sensing images. Existing contrastive learning methods encourage invariant feature learning by bringing positive samples defined by random transformations in feature spaces closer, where transformed samples of the same image at different intensities are considered equivalent. However, remote sensing images differ from natural images in their top-down perspective results in the arbitrary orientation of objects and in that the images contain rich in-plane rotation information. Maintaining invariance to rotation transformations can lead to the loss of rotation information in features, thereby affecting angle information predictions for differently rotated samples in downstream tasks. Therefore, we believe that contrastive learning should not focus only on strict invariance but encourage features to be equivariant to rotation while maintaining invariance to other transformations. To achieve this goal, we propose an invariant–equivariant covariant network (Co-ECL) based on collaborative and reverse mechanisms. The collaborative mechanism encourages rotation equivariance by predicting the rotation transformations of input images and combines invariant and equivariant learning tasks to jointly supervise the feature learning process to achieve collaborative learning. The reverse mechanism introduces a reverse rotation module in the feature learning stage, applying reverse rotation transformations with equal intensity to features in invariant learning tasks as in the data transformation stage, thereby ensuring their independent realization. In experiments conducted on three publicly available oriented object detection datasets of remote sensing images, our method consistently demonstrated the best performance. Additionally, these experiments on multi-angle datasets demonstrated that our method has good robustness on rotation-related tasks.

Forest stock volume is the main factor to evaluate forest carbon sink level. At present, the combination of multi-source remote sensing and non-parametric models has been widely used in FSV estimation. However, the biodiversity of natural forests is complex, and the response of the spatial information of remote sensing images to FSV is significantly reduced, which seriously affects the accuracy of FSV estimation. To address this challenge, this paper takes China’s Baishanzu Forest Park with representative characteristics of natural forests as the research object, integrates the forest survey data, SRTM data, and Landsat 8 images of Baishanzu Forest Park, constructs a time series dataset based on survey time, and establishes an FSV estimation model based on the CNN-LSTM-Attention algorithm. The model uses the convolutional neural network to extract the spatial features of remote sensing images, uses the LSTM to capture the time-varying characteristics of FSV, captures the feature variables with a high response to FSV through the attention mechanism, and finally completes the prediction of FSV. The experimental results show that some features (e.g., texture, elevation, etc.) of the dataset based on multi-source data feature variables are more effective in FSV estimation than spectral features. Compared with the existing models such as MLR and RF, the proposed model achieved higher accuracy in the study area (R2 = 0.8463, rMSE = 26.73 m3/ha, MAE = 16.47 m3/ha).

The Kravarsko area is located in a hilly region of northern Croatia, where numerous landslides endanger and damage houses, roads, water systems, and power lines. Nevertheless, natural hazard management plans are practically non-existent. Therefore, during the initial research, a landslide inventory was developed for the Kravarsko pilot area based on remote sensing data (high-resolution digital elevation models), and some of the landslides were investigated in detail. However, due to the complexity and vulnerability of the area, additional zoning of landslide-susceptible areas was needed. As a result, a slope gradient map, a map of engineering geological units, and a land-cover map were developed as inputs for the landslide susceptibility map. Additionally, based on the available data and a landslide inventory, a terrain stability map was developed for landslide management. Analysis and map development were performed within a geographical information system environment, and the terrain stability map with key infrastructure data was determined to be the “most user-friendly and practically usable” resource for non-expert users in natural hazard management, for example, the local administration. At the same time, the terrain stability map can easily provide practical information for the local community and population about the expected landslide “risk” depending on the location of infrastructure, estates, or objects of interest or for the purposes of future planning.

Ecological sustainability has recently risen to prominence in scientific research and management applications. Approaches to measuring ecological connectivity and their application to optimize ecological network (EN) design are powerful tools against landscape fragmentation and biodiversity loss.We focused on building an EN by identifying the most sensitive areas for ecological connectivity within the Reggio Calabria (Italy) metropolitan area. We also proposed a defragmentation scenario to improve the obtained EN.The CORINE Land Cover and the Urban Atlas 2018 were used to obtain a fine-scale representation of the study area. Ten terrestrial mammal species were used to model connectivity following a multi-species approach. Dispersal distance, patch size, and resistance to species movement were used to identify patches and corridors. Vegetational fractional coverage based on three years time series of Sentinel-2 red-edge normalized difference vegetation index was used to discriminate areas with higher naturalness. We used graph theory and connectivity metrics to test the EN’s robustness and identify locations for restoration in a defragmentation scenario.The obtained EN, formed by three separate components, was composed of 724 arcs and 300 nodes with an average patch area of 27.04 ha. After the defragmentation hypothesis, the EN, formed by only one component, was composed of 771 arcs and 328 nodes with an average patch area of 26.82 ha.It was possible to analyze an EN’s connectivity and evaluate the impact of a scenario intended to enhance multi-species connectivity. By comparing several connectivity metrics, we highlighted the potential of land interventions as a planning tool to enhance future ecological sustainability and biodiversity conservation.

With the remarkable success of change detection (CD) in remote sensing images in the context of deep learning, many convolutional neural network (CNN) based methods have been proposed. In the current research, to obtain a better context modeling method for remote sensing images and to capture more spatiotemporal characteristics, several attention-based methods and transformer (TR)-based methods have been proposed. Recent research has also continued to innovate on TR-based methods, and many new methods have been proposed. Most of them require a huge number of calculation to achieve good results. Therefore, using the TR-based mehtod while maintaining the overhead low is a problem to be solved. Here, we propose a GNN-based multi-scale transformer siamese network for remote sensing image change detection (GMTS) that maintains a low network overhead while effectively modeling context in the spatiotemporal domain. We also design a novel hybrid backbone to extract features. Compared with the current CNN backbone, our backbone network has a lower overhead and achieves better results. Further, we use high/low frequency (HiLo) attention to extract more detailed local features and the multi-scale pooling pyramid transformer (MPPT) module to focus on more global features respectively. Finally, we leverage the context modeling capabilities of TR in the spatiotemporal domain to optimize the extracted features. We have a relatively low number of parameters compared to that required by current TR-based methods and achieve a good effect improvement, which provides a good balance between efficiency and performance. [ABSTRACT FROM AUTHOR]

Urban sprawl is known to have a negative impact on the environment and is considered one of the main challenges in urban development. The current work aims to understand the spatio-temporal characteristics of urban growth and its environmental impact in the city of Amman. The paper's empirical method relies on NDVI, NDBI, and LST calculated using remote sensing and GIS techniques. For this study, Landsat Thematic Mapper Landsat 5-TM images from 1990, 1995, 1999, and 2004 were acquired, as well as Landsat 8-OLI images from 2013, 2017, and 2022. The surface temperature of Amman has increased as the area of paved roads, residential, commercial and industrial land use types have increased, while green spaces and vacant lots have decreased. The NDVI analysis revealed that there was variation in the vegetation index during the study year due to human activities as well as climatic change. Correlations between biophysical variables and LST revealed that NDVI had a significant negative correlation, while NDBI had a positive relationship. Further, a significant positive correlation was observed between road networks and built-up areas which enhance the occurrence of urban sprawl due to urbanization. This study highlighted the significance of considering the environmental consequences of urban sprawl when developing and implementing GAM policies and strategies. [ABSTRACT FROM AUTHOR]

In this work, remote sensing (RS) and Geographical Information System (GIS) techniques were developed to update drainage and surface water bodies as well as assess relief, linear, and aerial morphometric characteristics of the Gomti river watershed in Uttar Pradesh. The Gomti River Basin covers 3520 square kilometers. According to morphometric research, the drainage networks of the Gomti river basin have a sub-parallel drainage pattern and are subdendritic. Because of variations in slope and terrain, the stream length ratio varies. The study concluded that the streams in the Gomti River Basin watershed are at a mature stage, indicating geomorphic development. The topography and geometric growth of the watershed are used to explain differences in bifurcation ratio values. Stream frequencies in watersheds have a positive relationship with drainage density, indicating an increase in stream population as drainage density rises. The goal of this study is to examine the morphometric analysis of the Gomti river watershed under various physiographic situations. One of the most crucial analyses for the management and development of watersheds is morphometric analysis. [ABSTRACT FROM AUTHOR]

Learning long-range contextual dependence is important for remote sensing (RS) image segmentation in complex patterns. Meanwhile, exploring local context information is conducive to the discrimination of fine details. Only underlining either global semantic correlations or local context details is insufficient to achieve accurate segmentation. In this letter, we propose an architecture with the global-local self-attention (GLSA) mechanism, called GLSANet, which can simultaneously consider both global and local contexts for segmentations. Particularly, the GLSA mechanism consists of the global atrous self-attention (GASA) and local window self-attention (LWSA) mechanisms. GASA can learn long-range semantic relations in a gapped manner, while LWSA can locally capture contextual details. As a bridge between the two self-attention (SA) branches, a context fusion module (CFM) is further designed to adaptively integrate global and local contexts. The experiments with public datasets show that the proposed GLSANet significantly refines semantic segmentation and outperforms other competing methods. [ABSTRACT FROM AUTHOR]

Remote sensing (RS) object detectors based on convolutional neural networks (CNNs) are hard to model global context dependencies. Transformer-based detectors can overcome this problem via global pairwise interactions, but more comprehensive information interactions are not systemically investigated to boost the detection performance. In view of this issue, we propose a transformer-guided multi-interaction network (TransMIN), which uses ResNet50-feature pyramid network (FPN) as the backbone for remote sensing object detection (RSOD). Specifically, we implement local–global feature interactions (LGFIs) by combining convolution and transformer in the residual blocks of ResNet50 to learn complementary features. We implement cross-view feature interactions (CVFIs) via transformers in the pyramid layers of FPN to capture the correlation between reference features (spatial edge priors and channel statistics) and pyramid features. This enhances edge information and suppresses background interference. In the detection head, we adopt a task-interactive sample assigner (TISA) by considering the interactions of classification and localization losses to obtain high-quality predictions. Experiments on two benchmark datasets demonstrate the superior detection performance of TransMIN over state-of-the-art methods. [ABSTRACT FROM AUTHOR]

General convolutional neural networks (CNNs) process the data in the spatial domain and lack efficacy in modeling the spectral context in multispectral image processing task. To address this limitation, in this letter, we propose a wavelet-inspired attention-based convolution neural network (WIANet) architecture that combines the modeling of the spatio-spectral resolution of multispectral remote sensing (RS) images with a wavelet convolution and attention unit into a single deep learning architecture for land cover classification. Our aim is to add the characteristics of both wavelet transform (WT) and attention mechanism into an UNet-based architecture to better exploit the spectral and texture information for distinguishing classes with high similarity in the spectral signatures. We evaluate the performance of the proposed approach on a multilabel multispectral Sentinel-2 dataset. The experiment shows that the proposed approach performs better than the reference methods in different conditions on the number of training samples. [ABSTRACT FROM AUTHOR]