Single Synthetic Aperture Radar Research Articles

Assessing the Potential of Multi-Temporal Conditional Generative Adversarial Networks in SAR-to-Optical Image Translation for Early-Stage Crop Monitoring

The incomplete construction of optical image time series caused by cloud contamination is one of the major limitations facing the application of optical satellite images in crop monitoring. Thus, the construction of a complete optical image time series via image reconstruction of cloud-contaminated regions is essential for thematic mapping in croplands. This study investigates the potential of multi-temporal conditional generative adversarial networks (MTcGANs) that use a single synthetic aperture radar (SAR) image acquired on a prediction date and a pair of SAR and optical images acquired on a reference date in the context of early-stage crop monitoring. MTcGAN has an advantage over conventional SAR-to-optical image translation methods as it allows input data of various compositions. As the prediction performance of MTcGAN depends on the input data composition, the variations in the prediction performance should be assessed for different input data combination cases. Such an assessment was performed through experiments using Sentinel-1 and -2 images acquired in the US Corn Belt. MTcGAN outperformed existing SAR-to-optical image translation methods, including Pix2Pix and supervised CycleGAN (S-CycleGAN), in cases representing various input compositions. In particular, MTcGAN was substantially superior when there was little change in crop vitality between the reference and prediction dates. For the SWIR1 band, the root mean square error of MTcGAN (0.021) for corn was significantly improved by 54.4% and 50.0% compared to Pix2Pix (0.046) and S-CycleGAN (0.042), respectively. Even when there were large changes in crop vitality, the prediction accuracy of MTcGAN was more than twice that of Pix2Pix and S-CycleGAN. Without considering the temporal intervals between input image acquisition dates, MTcGAN was found to be beneficial when crops were visually distinct in both SAR and optical images. These experimental results demonstrate the potential of MTcGAN in SAR-to-optical image translation for crop monitoring during the early growth stage and can serve as a guideline for selecting appropriate input images for MTcGAN.

A GAN-Based Augmentation Scheme for SAR Deceptive Jamming Templates with Shadows

To realize fast and effective synthetic aperture radar (SAR) deception jamming, a high-quality SAR deception jamming template library can be generated by performing sample augmentation on SAR deception jamming templates. However, the current sample augmentation schemes of SAR deception jamming templates face certain problems. First, the authenticity of the templates is low due to the lack of speckle noise. Second, the generated templates have a low similarity to the target and shadow areas of the input templates. To solve these problems, this study proposed a sample augmentation scheme based on generative adversarial networks, which can generate a high-quality library of SAR deception jamming templates with shadows. The proposed scheme solved the two aforementioned problems from the following aspects. First, the influence of the speckle noise was considered in the network to avoid the problem of reduced authenticity in the generated images. Second, a channel attention mechanism module was used to improve the network’s learning ability of the shadow features, which improved the similarity between the generated template and the shadow area in the input template. Finally, the single generative adversarial network (SinGAN) scheme, which is a generative adversarial network capable of image sample augmentation for a single SAR image, and the proposed scheme were compared regarding the equivalent number of looks and the structural similarity between the target and shadow in the sample augmentation results. The comparison results demonstrated that, compared to the templates generated by the SinGAN scheme, those generated by the proposed scheme had targets and shadow features similar to those of the original image and could incorporate speckle noise characteristics, resulting in a higher authenticity, which helps to achieve fast and effective SAR deception jamming.

Application of different internal solitary wave theories for SAR remote sensing inversion in the northern South China Sea

This study discusses the theoretical method of the inversion of internal solitary waves (ISWs) characteristic parameters using synthetic aperture radar (SAR) images and ocean reanalysis data. Based on the horizontal distribution characteristics of the pixel intensity of the internal wave fringe in the image, an imaging theoretical model based on the extended Korteweg-de Vries (eKdV) and Benjamin-Ono (BO) theories in SAR was constructed, and the peak-to-peak method and curve-fitting method for calculating the characteristic wavelength of internal solitary waves were determined. The results show that the eKdV theoretical model has the advantage of inverting solitary waves with large amplitudes under the same marine stratified structure, and it provides a more reasonable characteristic parameter estimation than the traditional Korteweg-de Vries (KdV) theory. Although the BO theory is theoretically applicable to deep-water internal waves whose wavelength is far less than the water depth, it can still provide good inversion results in practical applications. Compared to the peak-to-peak method, the curve fitting method fully considers the pixel intensity variation characteristics of the internal wave fringe profile of the image, and the calculated characteristic parameters are more reasonable. Using the maximum buoyancy frequency and internal wave vertical mode to construct the marine stratification model of the deep waters of the northeastern South China Sea may have great uncertainty, and it is better to calculate the interface depth using the relative error of the coefficients of two-layer and continuous internal wave equations. In addition, this study also proposes an internal wave inversion method using continuous SAR images, which provides more reference and support for the selection of internal wave inversion models and reliability test of inversion results in a single SAR image.

A Geometry-Aware Consistent Constraint for Height Estimation From a Single SAR Imagery in Mountain Areas

Height estimation from a single synthetic aperture radar(SAR) imagery has shown a great potential in real-time scene understanding and environment detection. It is a mathematical ill-posed problem for that a single 2D image may be projected from multiple 3D scenes. Then, the problem is that the accuracy of height estimation from a single SAR image is not high enough without prior knowledge, especially in mountain areas. Thus, we proposes a geometry-aware consistent constraint(GACC) to improve the performance of single SAR height estimation. The simulated SAR images generated from height maps are used to establish high precision transformation between map and radar coordinates in mountain areas. The main idea of GACC is that the simulated SAR image generated from estimated height map should be consistent with the ground truth simulated SAR image. A sparse height map is included as the supplementary inputs to further improve the accuracy of estimated height maps. Comparison experiments are performed in Geermu, Huangshan and Guiyang datasets and the results show that the RMSE of the estimated height map by Unet with GACC and 0.0434% sparse height gets reduced by about 98%, 78% and 94% compared with that by Unet in the three datasets.

DDFormer: A Dual-Domain Transformer for Building Damage Detection Using High-Resolution SAR Imagery

Earthquakes are catastrophic in terms of damage to buildings. Synthetic aperture radar (SAR) has emerged as an effective tool to respond to seismic hazards. However, pre- and post-event high-resolution data are not always available for the affected areas, and the complex geometric properties of buildings pose a challenge to building damage detection. Therefore, this letter proposes the Dual-Domain Transformer (DDFormer) semantic segmentation model for damaged buildings detection using a single post-earthquake high-resolution SAR image. The difference between intact and collapsed building features is enhanced by adaptive frequency and spatial modules. Taking the 2023 Turkey earthquake as an example, the experiments are conducted on two high-resolution co-polarized SAR data (Capella and GF-3). The DDFormer achieves optimal detection accuracy with mean IOU (mIOU) and F-Score of 81.81% and 90%, respectively. In addition, our results are in high consistent with the Turkey Earthquake Report published by Microsoft with a correlation coefficient of 0.626. The above experiments demonstrate the robustness and effectiveness of DDFormer.

WCDL: A Weighted Cloud Dictionary Learning Method for Fusing Cloud-Contaminated Optical and SAR Images

Cloud cover hinders accurate and timely monitoring of urban land cover (ULC). The combination of synthetic aperture radar (SAR) and optical data without cloud contamination has demonstrated promising performance in previous research. However, ULC studies on cloud-prone areas are scarce despite the inevitability of cloud cover, especially in the tropics and subtropics. This study proposes a novel weighted cloud dictionary learning method (WCDL) for fusing optical and SAR data for the ULC classification in cloud-prone areas. We innovatively propose a cloud probability weighting model and a pixel-wise cloud dictionary learning method that take the interference disparities at various cloud probability levels into account to mitigate cloud interference. Experiments reveal that the overall accuracy (OA) of fused data rises by more than 6% and 20% compared to single SAR and optical data, respectively. This method considerably improved by 3% in OA compared with other methods that directly stitch optical and SAR data together regardless of cloud interference. It improves almost all land covers producer's accuracy (PA) and user's accuracy (UA) by up to 9%. Ablation studies further show the cloud probability weighting model improves the OA of all classifiers by up to 5%. And the pixel-wise cloud dictionary learning model improves by more than 2% in OA for all cloud conditions, and the UA and PA are enhanced by up to 9% and 10%. The proposed WCDL method will serve as a reference for fusing cloud-contaminated optical and SAR data and timely, continuous, and accurate land surface monitoring in cloudy areas.

TransUNet++SAR: Change Detection with Deep Learning about Architectural Ensemble in SAR Images

In the application of change detection satellite remote sensing images, synthetic aperture radar (SAR) images have become a more important data source. This paper proposes a new end-to-end SAR image change network architecture—TransUNet++SAR—that combines Transformer with UNet++. First, the convolutional neural network (CNN) was used to obtain the feature maps of the single time SAR images layer by layer. Tokenized image patches were encoded to extract rich global context information. Using improved Transformer for effective modeling of global semantic relations can generate rich contextual feature representations. Then, we used the decoder to upsample the encoded features, connected the encoded multi-scale features with the high-level features by sequential connection to learn the local-global semantic features, recovered the full spatial resolution of the feature map, and achieved accurate localization. In the UNet++ structure, the bitemporal SAR images are composed of two single networks, which have shared weights to learn the features of the single temporal image layer by layer to avoid the influence of SAR image noise and pseudo-change on the deep learning process. The experiment results show that the experimental effect of TransUNet++SAR on the Beijing, Guangzhou, and Qingdao datasets were significantly better than other deep learning SAR image change detection algorithms. At the same time, compared with other Transformer related change detection algorithms, the description of the changed area edge was more accurate. In the dataset experiments, the model had higher indices than the other models, especially the Beijing building change datasets, where the IOU was 9.79% higher and F1-score was 4.38% higher.

SAR2HEIGHT: Height Estimation from a Single SAR Image in Mountain Areas via Sparse Height and Proxyless Depth-Aware Penalty Neural Architecture Search for Unet

Height estimation from a single Synthetic Aperture Radar (SAR) image has demonstrated a great potential in real-time environmental monitoring and scene understanding. The projection of a single 2D SAR image from multiple 3D height maps is an ill-posed problem in mathematics. Although Unet has been widely used for height estimation from a single image, the ill-posed problem cannot be completely resolved, and it leads to deteriorated performance with limited training data. This paper tackles the problem by Unet with the help of supplementary sparse height information and proxyless neural architecture search (PDPNAS) for Unet. The sparse height, which can be accepted from low-resolution SRTM or LiDAR products, is included as the supplementary information and is helpful to improve the accuracy of the estimated height map, especially in mountain areas with a wide range of elevations. In order to explore the effect of sparsity of sparse height on the estimated height map, a parameterized method is proposed to generate sparse height with a different sparse ratio. In order to further improve the accuracy of the estimated height map from a single SAR imagery, PDPNAS for Unet is proposed. The optimal architecture for Unet can be searched by PDPNAS automatically with the help of a depth-aware penalty term p. The effectiveness of our approach is evaluated by visual and quantitative analysis on three datasets from mountain areas. The root mean squared error (RMSE) is reduced by 90.30% through observing only 0.0109% of height values from a low-resolution SRTM product. Furthermore, the RMSE is reduced by 3.79% via PDPNAS for Unet. The research proposes a reliable method for estimating height and an alternative method for wide-area DEM mapping from a single SAR image, especially for the implementation of real-time DEM estimation in mountain areas.

Shallow Sea Topography Detection from Multi-Source SAR Satellites: A Case Study of Dazhou Island in China

Accurate measurement of underwater topography in the coastal zone is essential for human marine activities, and the synthetic aperture radar (SAR) presents a completely new solution. However, underwater topography detection using a single SAR image is vulnerable to the interference of sea state and sensor noise, which reduces the detection accuracy. A new underwater topography detection method based on multi-source SAR (MSSTD) was proposed in this study to improve the detection precision. GF-3, Sentinel-1, ALOS PALSAR, and ENVISAT ASAR data were used to verify the sea area of Dazhou Island. The detection result was in good agreement with the chart data (MAE of 2.9 m and correlation coefficient of 0.93), and the detection accuracy was improved over that of a single SAR image. GF-3 image with 3 m spatial resolution performed best in bathymetry among the four SAR images. Additionally, the resolution of the SAR image had greater influence on bathymetry compared with polarization and radar band. The ability of MSSTD has been proved in our work. Collaborative multi-source satellite observation is a feasible and effective scheme in marine research, but its application potential in underwater topography detection still requires further exploration.

A Comprehensive Comparison of Machine Learning and Feature Selection Methods for Maize Biomass Estimation Using Sentinel-1 SAR, Sentinel-2 Vegetation Indices, and Biophysical Variables

Rapid and accurate estimation of maize biomass is critical for predicting crop productivity. The launched Sentinel-1 (S-1) synthetic aperture radar (SAR) and Sentinel-2 (S-2) missions offer a new opportunity to map biomass. The selection of appropriate response variables is crucial for improving the accuracy of biomass estimation. We developed models from SAR polarization indices, vegetation indices (VIs), and biophysical variables (BPVs) based on gaussian process regression (GPR) and random forest (RF) with feature optimization to retrieve maize biomass in Changchun, Jilin province, Northeastern China. Three new predictors from each type of remote sensing data were proposed based on the correlations to biomass measured in June, July, and August 2018. The results showed that a predictor combined by vertical-horizontal polarization (VV), vertical-horizontal polarization (VH), and the difference of VH and VV (VH-VV) derived from S-1 images of June, July, and August, respectively, with GPR and RF, provided a more accurate estimation of biomass (R2 = 0.81–0.83, RMSE = 0.40–0.41 kg/m2) than the models based on single SAR polarization indices or their combinations, or optimized features (R2 = 0.04–0.39, RMSE = 0.84–1.08 kg/m2). Among the S-2 VIs, the GPR model using a combination of ratio vegetation index (RVI) of June, normalized different infrared index (NDII) of July, and normalized difference vegetation index (NDVI) of August achieved a result with R2 = 0.83 and RMSE = 0.39 kg/m2, much better than single VIs or their combination, or optimized features (R2 of 0.31–0.77, RMSE of 0.47–0.87 kg/m2). A BPV predictor, combined with leaf chlorophyll content (CAB) in June, canopy water content (CWC) in July, and fractional vegetation cover (FCOVER) in August, with RF, also yielded the highest accuracy (R2 = 0.85, RMSE = 0.38 kg/m2) compared to that of single BPVs or their combinations, or optimized subset. Overall, the three combined predictors were found to be significant contributors to improving the estimation accuracy of biomass with GPR and RF methods. This study clearly sheds new insights on the application of S-1 and S-2 data on maize biomass modeling.

Improving Forest Above-Ground Biomass Retrieval Using Multi-Sensor L- and C- Band SAR Data and Multi-Temporal Spaceborne LiDAR Data

Due to the great structural and species diversity of tropical forests and limitations of the methods used to estimate aboveground biomass, there is uncertainty in quantifying its carbon sequestration potential. Measuring carbon sequestered in the terrestrial ecosystem and monitoring its dynamics is one of the key objectives in sustainable development goals. Synthetic Aperture Radar (SAR) has evolved as a key satellite technology in measuring and monitoring terrestrial carbon sink stored as biomass in plants. This study attempts to model forest above-ground biomass (AGB) using a random forest machine-learning approach where the predictor variables are from C-band (Radarsat-2), L-band (ALOS-2/PALSAR-2), and multi-temporal spaceborne LiDAR data from the GEDI platform. Training and validation data for the machine learning approach are obtained from the field measured inventory campaigns to evaluate the modeled forest biomass accuracies. The results show that variables from L-band (HH, HV), C-band (HV), and canopy height from the GEDI LiDAR platform performed effectively to model forest AGB with the coefficient of determination (R2) of 0.81 and root mean squared error (rmse) of 19.35 Mg/ha (%rmse – 17.17). In the case of single frequency SAR data, the analysis shows that the model derived from the L-band SAR data and LiDAR performed comparably better than the combination of C-band SAR and LiDAR data with an R2 of 0.78 and rmse of 21.36 Mg/ha (%rmse – 18.94). The results, thus, demonstrate the potential of SAR data (both single frequency and multiple frequencies) in combination with GEDI LiDAR data in effectively modeling AGB over highly biodiverse tropical forest regions.

Single Channel SAR Ground Moving Target Detection Algorithm Based on Subband Interferometry

Traditional synthetic aperture radar (SAR) ground moving target detection (GMTD) technology mostly relies on multichannel data, which requires higher complexity of the radar system. This letter proposes a subband interferometry (SBI)-based algorithm for single-channel SAR GMTD. First, the echo signal is divided into high and low bands containing overlapping parts before the imaging process, then the subbands are imaged separately, and the SBI algorithm is used to achieve the interferometric phase of the subbands. It can be observed that the moving target has the characteristics of phase reversal in the interferometric phase image, and graphics operations such as median filtering or morphological processing can be used to detect the moving target. In this letter, the relationship between the subbands coincidence, the correlation coefficient (CC) of SBI, and the phase of moving target after SBI processing are derived. The real airborne SAR data verify the feasibility and effectiveness of the algorithm.

End-to-End Method With Transformer for 3-D Detection of Oil Tank From Single SAR Image

In recent years, deep learning has been successfully applied in the field of synthetic aperture radar (SAR) image object detection. However, unlike ships and tanks targets, the oil tank targets in SAR image are usually dense and compact with more overlaps and discrete scattering centers, which greatly increases the difficulty of extracting location and structural parameters. Most of the existing methods transfer the methods suitable for natural image to the SAR image field, without considering the unique characteristics of SAR image. Therefore, in this article, we propose an improved model based on the end-to-end transformer network, which is the first model introducing transformer network to 3-D detection of oil tank targets from single SAR image. We input the incidence angle into the transformer model as a priori token. Then, we propose a feature description operator (FDO) based on the scattering centers that are used as an aid to improve the precision of predictions. In addition, we also propose a cylinder IOU as a more suitable evaluation metric for 3-D detection of oil tank. Finally, we evaluate our model on an SAR image dataset that contains SAR images from RADARSAT-2, TerraSAR-X, and GF-3 with different incidence angles. Our experiments demonstrate that our proposed model achieves the AP of 77.6% compared with 60.8% of baseline, which proves the effectiveness of the introduction of observation conditions, cylinder IOU loss (CI Loss), and the FDO based on the scattering centers in our model and is appealing for 3-D detection of oil tank.

Evaluation and Improvement of Generalization Performance of SAR Ship Recognition Algorithms

As artificial intelligence continues to advance, deep learning has greatly contributed to the advancement of ship recognition using Synthetic Aperture Radar (SAR) images. Deep learning-based SAR ship recognition performance is largely dependent on the sample set used. SAR ship recognition datasets published in recent years, however, are most derived from a single SAR satellite sensor. It needs to be evaluated and analyzed carefully whether the model trained by a single satellite dataset can still achieve the same accuracy with different SAR satellites. The paper focuses on the following research to address these issues. Firstly, using multiple SAR satellite sensors, we create a new SAR ship dataset (named generalization performance evaluation dataset, GPED) containing multi-resolution and multi-polarization data to examine the generalization performance of the deep learning-based SAR ship recognition method. GPED and a marine target detection dataset (MTDD) are then used to evaluate and analyze the generalization performance of current mainstream deep learning methods. According to the the experiment results, the mean average accuracy (mAP) of the ship recognition model trained on GPED is generally higher than that of MTDD, which proves that GPED has a better generalization performance. Furthermore, SAR ship detection datasets have more samples than ship recognition datasets, which inspired us to use transfer learning to transfer knowledge from ship detection to ship recognition. In this paper, a method for ship recognition based on transfer learning that utilizes the knowledge gained from the ship detection task is proposed. The method includes two modules: pretraining module and fine-tuning module. It can apply samples of unlabeled ship types to ship recognition, thus reducing the number of labeled samples that are required for ship recognition. The experimental results on GPED and MTDD show that our method can achieve good recognition performances.

Autofocus-Based Estimation of Penetration Depth and Permittivity of Ice Volumes and Snow Using Single SAR Images

An intrinsic challenge in the geophysical interpretation of low-frequency synthetic aperture radar (SAR) imagery of semitransparent media, such as ice sheets, is the position ambiguity of the scattering structures within the glacial volume. Commonly tackled by applying interferometric and tomographic techniques, their spaceborne implementation exhibits by orders higher complexity compared to missions relying on single SAR images, making them cost expensive or, in the context of planetary missions, even impossible due to limited navigation capability. Besides, even these sophisticated techniques are commonly biased due to inaccurate permittivity estimates, leading to geometric distortions up to several meters. We present a novel inversion procedure to estimate volume parameters of ice sheets, namely, the depth of the scattering layer within the glacial volume and the dielectric permittivity of the ice, based on single-image single-polarization SAR acquisitions. The information is inherent in the processed SAR data as phase errors on the azimuth signals resulting from uncompensated nonlinear propagation of the radar echoes through ice. We suggest a local map-drift autofocus approach to quantify and spatially resolve the phase errors and an inversion model to relate them to the penetration depth and permittivity. Testing the proposed technique using P-band SAR data acquired using DLR’s airborne sensor F-SAR during the ARCTIC15 campaign in Greenland shows promising results and good agreement with tomographic products of the analyzed test site.

Assessing the Impact of Positioning Errors in Car-Borne Repeat-Pass SAR Interferometry With a Controlled Rail-Based Experiment

Agile synthetic aperture radar (SAR) platforms such as car-borne and UAV-borne SAR systems require combined inertial navigation systems (INS) and global navigation satellite systems (GNSS) to measure the radar sensor trajectories used for focusing and interferometric processing. Measurement inaccuracies from INS/GNSS systems lead to residual phase errors in the SAR products whose minimisation is crucial to derive accurate topographic and deformation information. In this work, we analyse the impact of residual positioning errors on car-borne repeat-pass SAR interferometry at L-band for different INS/GNSS measurement configurations and for the typical car-borne acquisition geometry. The positioning errors are evaluated both during single SAR acquisitions with long integration times and between different acquisitions as a function of the distance of the radar platform from the GNSS reference stations. We show the reduction of interferometric phase errors achievable by additionally using a GNSS receiver mounted in the vicinity of the SAR platform as compared to remote reference stations of the national network of permanent GNSS receivers. Test results obtained in a controlled setup with a rail-based SAR system equipped with a navigation-grade INS/GNSS system show maximum repeat-pass trajectory errors on the order of 1–2 cm using a local GNSS reference station and up to 10–15 cm using the remote reference stations, leading to azimuth and range phase trends in the interferometric products.

A Fast Algorithm for High Accuracy Airborne SAR Geolocation Based on Local Linear Approximation

Synthetic aperture radar (SAR) is a kind of high-resolution imaging radar, which can work in any weather conditions and is often used for target positioning. Airborne SAR has the advantages of flexible scanning geometry and high spatial resolution compared with Spaceborne SAR, but the quality of SAR image and positioning accuracy is affected by air turbulence and mechanical vibration. In practical applications, it is vital to improve Airborne SAR’s geometric positioning accuracy without ground control points or other reference image information. When positioning a single SAR image with long-range, the Earth surface model is often used as an additional equation to solve the Range-Doppler (RD) model. However, the Earth surface equation is difficult to solve the target height accurately. This study proposed an Airborne SAR target positioning methodology based on a local linear approximation of the Earth surface equation. The proposed local Linear RD (L-RD) model establishes a linear constraint of the local elevation, and the iterative algorithm converges rapidly. To further improve the efficiency of geolocation, we propose a method to directly obtain the initial geolocation used for the iteration based on the geometric relationship. An iterative framework is also proposed to enhance positioning accuracy when DEM is available. The experimental results show that the proposed method can significantly improve the positioning accuracy and reliability of Airborne SAR positioning without control points compared to other methods.

Combined Estimation of Ionospheric Effects in SAR Images Exploiting Faraday Rotation and Autofocus

At lower frequencies, synthetic aperture radar (SAR) images can be significantly affected by ionospheric scintillations. This letter proposes an approach to estimate ionospheric artifacts in single SAR images by combining Faraday rotation (FR) estimates and autofocus methods by using a Wiener filter. This novel approach achieves superior performance compared to using these techniques independently. This improvement is demonstrated by injecting randomly simulated ionospheric phase screens into simulated point targets and airborne data adjusted to the parameters of the Biomass system.

Large-scale building height retrieval from single SAR imagery based on bounding box regression networks

Building height retrieval from synthetic aperture radar (SAR) imagery is of great importance for urban applications, yet highly challenging due to the complexity of SAR data. This paper addresses the issue of building height retrieval in large-scale urban areas from a single TerraSAR-X spotlight or stripmap image. Based on the radar viewing geometry, we propose that this problem be formulated as a bounding box regression problem and therefore allows for integrating height data from multiple data sources in generating ground truth on a larger scale. We introduce building footprints from geographic information system (GIS) data as complementary information and propose a bounding box regression network that exploits the location relationship between a building’s footprint and its bounding box, enabling fast computation. The method is validated on four urban data sets using TerraSAR-X images in both high-resolution spotlight and stripmap modes. Experimental results show that the proposed network can reduce the computation cost significantly while keeping the height accuracy of individual buildings compared to a Faster R-CNN based method. Moreover, we investigate the impact of inaccurate GIS data on our proposed network, and this study shows that the bounding box regression network is robust against positioning errors in GIS data. The proposed method has great potential to be applied to regional or even global scales. Our code will be made publicly available at github.com/ya0-sun/bbox-SAR-building.

A super resolution target separation and reconstruction approach for single channel sar against deceptive jamming

The excellent remote sensing ability of synthetic aperture radar (SAR) will be misled seriously when it encounters deceptive jamming which possesses high fidelity and fraudulence. In this paper, the dynamic synthetic aperture (DSA) scheme is used to extract the difference between the true and false targets. A simultaneous deceptive jamming suppression and target reconstruction method is proposed for a single channel SAR system to guarantee remote sensing ability. The system model is formulated as a sparse signal recovery problem with an unknown parametric dictionary to be estimated. An iterative re-weighted method is employed to jointly handle the dictionary parameter learning and target reconstruction problem in an majorization-minimization framework, where a surrogate function majorizing the Gaussian entropy in the objective function is introduced to circumvent its non-convexity. After dictionary parameter learning, the grid mismatching problem in a fixed grid based method is avoided. Therefore, the proposed method can reap a super resolution result. Besides, a simple yet effective DSA section scheme is developed for the SAR data excerpting, in which only two DSAs are required. Experimental results about location error and reconstruction power error reveal that the proposed method is able to achieve a good performance in deceptive jamming suppression.