Accepted Article Preview
More
Sparse signal processing-based Synthetic Aperture Radar (SAR) imaging, also known as sparse SAR imaging, is the main research direction of sparse microwave imaging theory. Compared with a conventional SAR system, sparse SAR imaging radar has significant potential to improve imaging performance. However, because it requires heavy computations, the application of sparse SAR imaging in large-scene recovery has become difficult, which restricts its further applications. Additionally, complex SAR images, rather than raw data, are usually used for data archiving due to a number of reasons such as data copyright and system confidentiality. Therefore, it is worthwhile to study how sparse imaging can be achieved using only Matched Filtering (MF)recovered complex images with less computational cost. GaoFen-3 is China’s first 1-m resolution multi-polarization C-band satellite. It has a high-resolution, wide swath imaging ability and hence plays an important role in disaster monitoring and ocean surveillance applications. In this paper, we introduce a complex image-based sparse SAR imaging method to process GaoFen-3 complex image data and improve image performance. Experimental results show that the sparse imaging results have lower sidelobes, higher signal-to-clutter and noise ratio, and better target distinguishing ability compared with inputted images. Additionally, sparse imaging can effectively preserve the statistical distribution and phase information of images that makes the recovered GaoFen-3 sparse image-based applications such as interferometric synthetic aperture radar and constant false alarm ratio detection possible. Sparse signal processing-based Synthetic Aperture Radar (SAR) imaging, also known as sparse SAR imaging, is the main research direction of sparse microwave imaging theory. Compared with a conventional SAR system, sparse SAR imaging radar has significant potential to improve imaging performance. However, because it requires heavy computations, the application of sparse SAR imaging in large-scene recovery has become difficult, which restricts its further applications. Additionally, complex SAR images, rather than raw data, are usually used for data archiving due to a number of reasons such as data copyright and system confidentiality. Therefore, it is worthwhile to study how sparse imaging can be achieved using only Matched Filtering (MF)recovered complex images with less computational cost. GaoFen-3 is China’s first 1-m resolution multi-polarization C-band satellite. It has a high-resolution, wide swath imaging ability and hence plays an important role in disaster monitoring and ocean surveillance applications. In this paper, we introduce a complex image-based sparse SAR imaging method to process GaoFen-3 complex image data and improve image performance. Experimental results show that the sparse imaging results have lower sidelobes, higher signal-to-clutter and noise ratio, and better target distinguishing ability compared with inputted images. Additionally, sparse imaging can effectively preserve the statistical distribution and phase information of images that makes the recovered GaoFen-3 sparse image-based applications such as interferometric synthetic aperture radar and constant false alarm ratio detection possible.
The special imaging mechanism of the Synthetic Aperture Radar (SAR) causes the sidelobe effect on SAR images. In target detection, the sidelobe effect changes the shapes of strong reflective targets, which results in the problems of localization difficulty and localization error. To solve this problem, this paper proposes a ship detection algorithm based on Spatially Variant Apodization (SVA) and Order Statistic-Constant False Alarm Rate (OS-CFAR). First, the global-CFAR algorithm is used to prescreen the potential target points, which reduces the computational burden of the following steps. Second, the SVA algorithm is modified to improve the speed of sidelobe suppression and applied to the raw complex image data. Then, the nonlinear method OS-CFAR is used to detect the targets on the processed image, and the morphological dilation processing is used to make up for the wrong suppressed points caused by the SVA algorithm. Finally, the GF-3 SAR images are used to test the algorithm and the comparison of the image contrast and detected numbers in the results with SVA and without SVA verifies the effectiveness of the proposed algorithm. The special imaging mechanism of the Synthetic Aperture Radar (SAR) causes the sidelobe effect on SAR images. In target detection, the sidelobe effect changes the shapes of strong reflective targets, which results in the problems of localization difficulty and localization error. To solve this problem, this paper proposes a ship detection algorithm based on Spatially Variant Apodization (SVA) and Order Statistic-Constant False Alarm Rate (OS-CFAR). First, the global-CFAR algorithm is used to prescreen the potential target points, which reduces the computational burden of the following steps. Second, the SVA algorithm is modified to improve the speed of sidelobe suppression and applied to the raw complex image data. Then, the nonlinear method OS-CFAR is used to detect the targets on the processed image, and the morphological dilation processing is used to make up for the wrong suppressed points caused by the SVA algorithm. Finally, the GF-3 SAR images are used to test the algorithm and the comparison of the image contrast and detected numbers in the results with SVA and without SVA verifies the effectiveness of the proposed algorithm.
Track initiation is the first important step in group target tracking, and it has a direct effect on the quality of the overall procedure. Traditional radar target tracking methods only utilize information about the target position to detect group numbers, but they do not use information relating to echo amplitude. Tracks are thus easily lost, as the numbers of detected groups and equivalent measurements are inaccurate. This paper proposes a group target track initiation method aided by echo amplitude information to ameliorate these problems. In this respect, target position and echo amplitude information is used to detect the number of target groups, and equivalent measurements are then computed using amplitude weighting and position weighting. Echo amplitude information is employed in the step of detecting group target numbers and computing the equivalent measurements, and group target tracks are subsequently initialized using the modified logic method. The proposed method can be used to correctly detect the number of target groups when the number is previously unknown. Furthermore, the method reduces the rate of track loss and improves the performance of group target tracking. The effectiveness of the proposed method is validated by the simulation results. Track initiation is the first important step in group target tracking, and it has a direct effect on the quality of the overall procedure. Traditional radar target tracking methods only utilize information about the target position to detect group numbers, but they do not use information relating to echo amplitude. Tracks are thus easily lost, as the numbers of detected groups and equivalent measurements are inaccurate. This paper proposes a group target track initiation method aided by echo amplitude information to ameliorate these problems. In this respect, target position and echo amplitude information is used to detect the number of target groups, and equivalent measurements are then computed using amplitude weighting and position weighting. Echo amplitude information is employed in the step of detecting group target numbers and computing the equivalent measurements, and group target tracks are subsequently initialized using the modified logic method. The proposed method can be used to correctly detect the number of target groups when the number is previously unknown. Furthermore, the method reduces the rate of track loss and improves the performance of group target tracking. The effectiveness of the proposed method is validated by the simulation results.
With the application of deep learning technology in the radar target recognition field, the automatic extraction of the target feature greatly improves the accuracy and robustness of the recognition, but its robustness in noisy environments needs to be further investigated. This paper proposes a robust target recognition method for radar High Resolution Range Profile (HRRP) data based on convolutional neural networks. By enhancing training set and using the residual block, inception structure, and denoising sparse autoencoder layer to enhance the network structure, a higher recognition rate is achieved in a wider SNR range, under the condition of 0 dB Rayleigh noise, the recognition rate reaches 96.14%, and the influence of the network structure and noise type on results is analyzed. With the application of deep learning technology in the radar target recognition field, the automatic extraction of the target feature greatly improves the accuracy and robustness of the recognition, but its robustness in noisy environments needs to be further investigated. This paper proposes a robust target recognition method for radar High Resolution Range Profile (HRRP) data based on convolutional neural networks. By enhancing training set and using the residual block, inception structure, and denoising sparse autoencoder layer to enhance the network structure, a higher recognition rate is achieved in a wider SNR range, under the condition of 0 dB Rayleigh noise, the recognition rate reaches 96.14%, and the influence of the network structure and noise type on results is analyzed.
Co-prime-sampling space-borne Synthetic Aperture Radar (SAR) replaces the traditional uniform sampling by performing co-prime sampling in azimuth, which effectively alleviates the conflict between spatial resolution and effective swath width, while also improving the ground detection performance of the SAR system. However, co-prime-sampling in azimuth causes the echo signal to exhibit azimuthal under sampling and non-uniform sampling characteristics, which means the traditional SAR image-processing method can not effectively image co-prime-sampled SAR. In this paper, an imaging method based on Two-Dimensional (2D) sparse-signal reconstruction is proposed for co-prime-sampling space-borne SAR. Using this method, after range-pulse compression, the 2D observed signal is intercepted and a corresponding sparse dictionary consisting of 2D atoms is constructed according to the Doppler parameters of each range gate. Then, azimuth-focus processing is completed by the improved 2D-signal sparsity adaptive matching pursuit algorithm. The proposed method not only compensates for the 2D coupling between the range and azimuth, but also eliminates the influence of space-varying imaging parameters on sparse reconstruction to achieve accurate reconstruction of the entire scene. The simulation results of the point targets and distribution targets verify that the proposed method can effectively reconstruct sparse scenes at a rate much lower than the Nyquist sampling rate. Co-prime-sampling space-borne Synthetic Aperture Radar (SAR) replaces the traditional uniform sampling by performing co-prime sampling in azimuth, which effectively alleviates the conflict between spatial resolution and effective swath width, while also improving the ground detection performance of the SAR system. However, co-prime-sampling in azimuth causes the echo signal to exhibit azimuthal under sampling and non-uniform sampling characteristics, which means the traditional SAR image-processing method can not effectively image co-prime-sampled SAR. In this paper, an imaging method based on Two-Dimensional (2D) sparse-signal reconstruction is proposed for co-prime-sampling space-borne SAR. Using this method, after range-pulse compression, the 2D observed signal is intercepted and a corresponding sparse dictionary consisting of 2D atoms is constructed according to the Doppler parameters of each range gate. Then, azimuth-focus processing is completed by the improved 2D-signal sparsity adaptive matching pursuit algorithm. The proposed method not only compensates for the 2D coupling between the range and azimuth, but also eliminates the influence of space-varying imaging parameters on sparse reconstruction to achieve accurate reconstruction of the entire scene. The simulation results of the point targets and distribution targets verify that the proposed method can effectively reconstruct sparse scenes at a rate much lower than the Nyquist sampling rate.
In the Synthetic Aperture Radar (SAR) remote sensing imagery of complicated scenes (especially urban scenes), there are a large number of lines and surfaces, such as roads in urban areas and the surfaces of buildings. These microwave-signal-scattering features have strong directivity. Traditional SAR acquires the scattering information of a scene from a single observation, and traditional imaging algorithms are based on the point target model, which makes the main features of the lines and surfaces in traditional SAR images appear as a series of strong scattering points rather than line-scattering and surface-scattering features. This outcome ultimately causes the target to be discontinuous in the SAR image, thus making the SAR image difficult to interpret. Therefore, in this study, we conducted an in-depth and meticulous investigation of the SAR imaging mechanism for lines and surfaces by establishing a parametric echo model of typical lines and triangular surfaces. Based on the proposed parametric model, we performed parametric imaging of these lines and surfaces. Based on our results, we propose a parametric imaging method, in which the typical lines and surfaces are classified and determined based on Bayesian theory and the proposed parametric model. Then, an SAR image can be obtained that effectively characterizes the scattering features of the line and surface targets by visual imaging, which effectively facilitates SAR image interpretation. The results of our numerical simulation experiments verify the validity of the proposed method. In the Synthetic Aperture Radar (SAR) remote sensing imagery of complicated scenes (especially urban scenes), there are a large number of lines and surfaces, such as roads in urban areas and the surfaces of buildings. These microwave-signal-scattering features have strong directivity. Traditional SAR acquires the scattering information of a scene from a single observation, and traditional imaging algorithms are based on the point target model, which makes the main features of the lines and surfaces in traditional SAR images appear as a series of strong scattering points rather than line-scattering and surface-scattering features. This outcome ultimately causes the target to be discontinuous in the SAR image, thus making the SAR image difficult to interpret. Therefore, in this study, we conducted an in-depth and meticulous investigation of the SAR imaging mechanism for lines and surfaces by establishing a parametric echo model of typical lines and triangular surfaces. Based on the proposed parametric model, we performed parametric imaging of these lines and surfaces. Based on our results, we propose a parametric imaging method, in which the typical lines and surfaces are classified and determined based on Bayesian theory and the proposed parametric model. Then, an SAR image can be obtained that effectively characterizes the scattering features of the line and surface targets by visual imaging, which effectively facilitates SAR image interpretation. The results of our numerical simulation experiments verify the validity of the proposed method.
Current Issue
More
Reviews
This paper analyzes the multi-directional evolution of radar ground imaging technology from the aspects of the representation of imaging results, aperture manifolds, signal channels, system morphologies, observation directions, processing methods, realization mechanisms, and target recognition. Attempts are made to analyze and understand the internal and external factors as well as the development law of radar ground imaging technology from a macroscopic perspective over a long time scale, and to predict the direction of future development. Alternative observation perspective and thinking method are proposed with a view to advance the understanding of the times veins and macro trends of radar ground imaging technology, meet practical needs, lead innovation efforts, and promote development and applications. This paper analyzes the multi-directional evolution of radar ground imaging technology from the aspects of the representation of imaging results, aperture manifolds, signal channels, system morphologies, observation directions, processing methods, realization mechanisms, and target recognition. Attempts are made to analyze and understand the internal and external factors as well as the development law of radar ground imaging technology from a macroscopic perspective over a long time scale, and to predict the direction of future development. Alternative observation perspective and thinking method are proposed with a view to advance the understanding of the times veins and macro trends of radar ground imaging technology, meet practical needs, lead innovation efforts, and promote development and applications.
Synthetic Aperture Radar three-Dimensional (SAR 3D) imaging technology can eliminate severe overlap in 2D images, and improve target recognition and 3D modeling capabilities, which have become an important trend in SAR development. After decades of development of SAR 3D imaging technology, many types of 3D imaging methods have been proposed. In this study, the history of SAR 3D imaging technology is systematically reviewed and the characteristics of existing SAR 3D imaging technology are analyzed. Given that the 3D information contained in SAR echo and images is not fully used by existing techniques, a new concept of SAR microwave vision 3D imaging has been proposed for the first time. This new concept is integrated with microwave scattering mechanism and image visual semantics to realize three-dimensional reconstruction, which form the theory and method of SAR microwave vision 3D imaging and can achieve high-efficiency and low-cost SAR 3D imaging. This study also analyzes the concept, goal and key scientific problems of SAR microwave vision 3D imaging and provides a preliminary solution, which will contribute in several ways to our understanding of SAR 3D imaging and provide the basis for further research. Synthetic Aperture Radar three-Dimensional (SAR 3D) imaging technology can eliminate severe overlap in 2D images, and improve target recognition and 3D modeling capabilities, which have become an important trend in SAR development. After decades of development of SAR 3D imaging technology, many types of 3D imaging methods have been proposed. In this study, the history of SAR 3D imaging technology is systematically reviewed and the characteristics of existing SAR 3D imaging technology are analyzed. Given that the 3D information contained in SAR echo and images is not fully used by existing techniques, a new concept of SAR microwave vision 3D imaging has been proposed for the first time. This new concept is integrated with microwave scattering mechanism and image visual semantics to realize three-dimensional reconstruction, which form the theory and method of SAR microwave vision 3D imaging and can achieve high-efficiency and low-cost SAR 3D imaging. This study also analyzes the concept, goal and key scientific problems of SAR microwave vision 3D imaging and provides a preliminary solution, which will contribute in several ways to our understanding of SAR 3D imaging and provide the basis for further research.
The development of multimode high-resolution Synthetic Aperture Radar (SAR) poses new challenges to information perception and feature abstraction of the space, ground, sea, and environment targets. The intersection of spatial remote sensing big data and artificial intelligence information technology is a new scientific research domain and major application area in Automatic Target Recognition (ATR). We emphasize that research on artificial intelligence information technology needs to be conducted under the physical background of the interaction between electromagnetic waves and targets, i.e., physical intelligence, to develop microwave vision of information perception on the electromagnetic spectrum that cannot be recognized by human eyes. This study is based on a keynote speech presented by author at the Fifth Young Scientists Forum of Journal of Radars on August 15, 2019. The development of multimode high-resolution Synthetic Aperture Radar (SAR) poses new challenges to information perception and feature abstraction of the space, ground, sea, and environment targets. The intersection of spatial remote sensing big data and artificial intelligence information technology is a new scientific research domain and major application area in Automatic Target Recognition (ATR). We emphasize that research on artificial intelligence information technology needs to be conducted under the physical background of the interaction between electromagnetic waves and targets, i.e., physical intelligence, to develop microwave vision of information perception on the electromagnetic spectrum that cannot be recognized by human eyes. This study is based on a keynote speech presented by author at the Fifth Young Scientists Forum of Journal of Radars on August 15, 2019.
As an important tool for acquiring remote sensing information, Synthetic Aperture Radar (SAR) has various modes, including high-resolution wide-swath, multi-angle information acquisition, high temporal observation, and three-dimensional topographic mapping. For any spaceborne SAR system, obtaining high-quality images is a prerequisite for improving the performance of SAR applications. In this paper, we analyze the factors affecting spaceborne SAR imaging and image quality with respect to orbit, platform, payload, and signal processing. We describe high-precision data acquisition techniques, including amplitude-phase compensation, the dynamic adjustment of the central electronic equipment, and antenna pattern estimation. We then present imaging compensation methods based on the improved motion model and tropospheric delay correction, which can achieve resolutions better than 0.3 m. Lastly, we summarize and compare SAR image processing techniques such as speckle noise suppression, azimuth ambiguity suppression, and sidelobe suppression, whereby the equivalent number of looks can be increased to more than 25 and the azimuth ambiguity and sidelobes can both be suppressed by 20 dB. As an important tool for acquiring remote sensing information, Synthetic Aperture Radar (SAR) has various modes, including high-resolution wide-swath, multi-angle information acquisition, high temporal observation, and three-dimensional topographic mapping. For any spaceborne SAR system, obtaining high-quality images is a prerequisite for improving the performance of SAR applications. In this paper, we analyze the factors affecting spaceborne SAR imaging and image quality with respect to orbit, platform, payload, and signal processing. We describe high-precision data acquisition techniques, including amplitude-phase compensation, the dynamic adjustment of the central electronic equipment, and antenna pattern estimation. We then present imaging compensation methods based on the improved motion model and tropospheric delay correction, which can achieve resolutions better than 0.3 m. Lastly, we summarize and compare SAR image processing techniques such as speckle noise suppression, azimuth ambiguity suppression, and sidelobe suppression, whereby the equivalent number of looks can be increased to more than 25 and the azimuth ambiguity and sidelobes can both be suppressed by 20 dB.
The multi-platform-borne Synthetic Aperture Radar (SAR) has become one of the most explored research directions in the domain of SAR. This study discusses the imaging algorithms in multi-platform-borne SARs such as airborne SAR, missile-borne SAR, and spaceborne SAR. First, the establishment of the radar echo model is briefly introduced, including two main points: slant range-model and imaging mode. Subsequently, the imaging algorithms of the aforementioned multi-platform-borne SARs developed and used in recent years are summarized. In addition, the inherent characteristics and challenges are described. Finally, the future development trends of the research are discussed. The multi-platform-borne Synthetic Aperture Radar (SAR) has become one of the most explored research directions in the domain of SAR. This study discusses the imaging algorithms in multi-platform-borne SARs such as airborne SAR, missile-borne SAR, and spaceborne SAR. First, the establishment of the radar echo model is briefly introduced, including two main points: slant range-model and imaging mode. Subsequently, the imaging algorithms of the aforementioned multi-platform-borne SARs developed and used in recent years are summarized. In addition, the inherent characteristics and challenges are described. Finally, the future development trends of the research are discussed.
“Contrast” is an generic denomination for “difference”. Measures of contrast are a powerful tool in image processing and analysis, e.g., in denoising, edge detection, segmentation, classification, parameter estimation, change detection, and feature selection. We present a survey on techniques that aim at measuring the contrast between (i) samples of SAR imagery, and (ii) samples and models, with emphasis on those that employ the statistical properties of the data. “Contrast” is an generic denomination for “difference”. Measures of contrast are a powerful tool in image processing and analysis, e.g., in denoising, edge detection, segmentation, classification, parameter estimation, change detection, and feature selection. We present a survey on techniques that aim at measuring the contrast between (i) samples of SAR imagery, and (ii) samples and models, with emphasis on those that employ the statistical properties of the data.
Papers
Spaceborne Synthetic Aperture Radar (SAR) is a type of microwave imaging radar with 2D high resolution. This technological device achieves range high resolution by transmitting wideband signals and azimuth high resolution through the synthetic aperture approach. With the increasing demand for high-resolution imaging, the resolution of spaceborne SAR has moved toward the decimeter level. On the one hand, limited by the present hardware technology, achieving wideband signal transmission through stepped-frequency technology is necessary. In this case, we need to study high-precision bandwidth synthesis technology. The influence of slant range error and amplitude and phase error between sub-bands should be considered. On the other hand, due to limited beamwidth, the system needs to work in sliding spot mode to achieve a long synthetic aperture. In this case, we need to study the problem of imaging parameter variance caused by curved orbit, “Stop–go” error, and the influence of ionospheric and tropospheric transmission errors on imaging. To solve these problems, this paper introduces the principle of stepped-frequency signal design and bandwidth synthesis technology in detail. A time-domain algorithm and non-ideal factor compensation method are proposed for spaceborne high-resolution stepped-frequency SAR imaging. Finally, simulation verification and performance analysis of the imaging algorithm are conducted. Spaceborne Synthetic Aperture Radar (SAR) is a type of microwave imaging radar with 2D high resolution. This technological device achieves range high resolution by transmitting wideband signals and azimuth high resolution through the synthetic aperture approach. With the increasing demand for high-resolution imaging, the resolution of spaceborne SAR has moved toward the decimeter level. On the one hand, limited by the present hardware technology, achieving wideband signal transmission through stepped-frequency technology is necessary. In this case, we need to study high-precision bandwidth synthesis technology. The influence of slant range error and amplitude and phase error between sub-bands should be considered. On the other hand, due to limited beamwidth, the system needs to work in sliding spot mode to achieve a long synthetic aperture. In this case, we need to study the problem of imaging parameter variance caused by curved orbit, “Stop–go” error, and the influence of ionospheric and tropospheric transmission errors on imaging. To solve these problems, this paper introduces the principle of stepped-frequency signal design and bandwidth synthesis technology in detail. A time-domain algorithm and non-ideal factor compensation method are proposed for spaceborne high-resolution stepped-frequency SAR imaging. Finally, simulation verification and performance analysis of the imaging algorithm are conducted.
Miniature Synthetic Aperture Radar (MiniSAR) has been making breakthroughs to effectively overcome the limitations of time and space, and has exhibited superiority with respect to light weight and low-power consumption as well as high flexibility to achieve high-resolution imaging for the Region Of Interest (ROI). However, imaging signal processing for MiniSAR systems still face several technical challenges such as high-resolution imaging of ground targets under complicated trajectories, refocusing of non-cooperative moving targets, and efficient and real-time processing of echo data. In this paper, a series of imaging signal processing and associated hardware designs using Field-Programmable Gate Array (FPGA) architecture have been proposed to realize MiniSAR ultra-high-resolution imaging and real-time processing. Additionally, experimental results utilizing considerable spotlight and stripmap MiniSAR data have been presented to demonstrate the effectiveness and reliability of the proposed technology. Miniature Synthetic Aperture Radar (MiniSAR) has been making breakthroughs to effectively overcome the limitations of time and space, and has exhibited superiority with respect to light weight and low-power consumption as well as high flexibility to achieve high-resolution imaging for the Region Of Interest (ROI). However, imaging signal processing for MiniSAR systems still face several technical challenges such as high-resolution imaging of ground targets under complicated trajectories, refocusing of non-cooperative moving targets, and efficient and real-time processing of echo data. In this paper, a series of imaging signal processing and associated hardware designs using Field-Programmable Gate Array (FPGA) architecture have been proposed to realize MiniSAR ultra-high-resolution imaging and real-time processing. Additionally, experimental results utilizing considerable spotlight and stripmap MiniSAR data have been presented to demonstrate the effectiveness and reliability of the proposed technology.
This paper describes the L-band airborne Synthetic Aperture Radar (SAR) system and an experimental study of repeat-pass SAR interferometry. The discussions mainly focus on the internal calibration method, the imaging method based on Differential Global Positioning System/Inertial Measurement Unit (DGPS/IMU) data, Residual Motion Error (RME) estimation and compensation, and the Digital Elevation Model (DEM) inversion method with Interferometric SAR (InSAR). Two typical test areas were chosen from the acquired SAR data for detailed analyses. The DEMs in these two areas were inversed with repeat-pass InSAR, and comparisons with the measured ground points show errors in these two areas with a standard deviation 3.40 m and 2.85 m, respectively. This demonstrates the good performance of this L-band repeat-pass interferometric SAR system. These results have important significance for the design of airborne repeat-pass InSAR systems and typical terrain mapping. When the flight tracks are under good control, this L-band SAR system is applicable in the fields of differential InSAR, polarimetric InSAR, and tomographic SAR. This paper describes the L-band airborne Synthetic Aperture Radar (SAR) system and an experimental study of repeat-pass SAR interferometry. The discussions mainly focus on the internal calibration method, the imaging method based on Differential Global Positioning System/Inertial Measurement Unit (DGPS/IMU) data, Residual Motion Error (RME) estimation and compensation, and the Digital Elevation Model (DEM) inversion method with Interferometric SAR (InSAR). Two typical test areas were chosen from the acquired SAR data for detailed analyses. The DEMs in these two areas were inversed with repeat-pass InSAR, and comparisons with the measured ground points show errors in these two areas with a standard deviation 3.40 m and 2.85 m, respectively. This demonstrates the good performance of this L-band repeat-pass interferometric SAR system. These results have important significance for the design of airborne repeat-pass InSAR systems and typical terrain mapping. When the flight tracks are under good control, this L-band SAR system is applicable in the fields of differential InSAR, polarimetric InSAR, and tomographic SAR.
The characteristics of airborne millimeter-wave Interferometric Synthetic Aperture Radar (InSAR) include unrestricted light, large surveying width, and high mapping precision. In recent years, with the continuous development and improvement of the technology of airborne millimeter-wave InSAR, it has gradually become a widely used mapping method. The core of the system design of a high-precision millimeter-wave InSAR system designed for small aircraft platforms comprises InSAR baseline configuration, multi-baseline configuration, the external Digital Elevation Model (DEM), and InSAR processing flow. In this study, interferometric elevation measurements influenced by different baseline parameters of an airborne millimeter-wave InSAR system are analyzed. A design scheme of the millimeter-wave multi-baseline InSAR system based on integrated antenna pod is provided. Then, a time-domain imaging algorithm-based millimeter-wave multi-baseline InSAR elevation measurement process is proposed. Finally, real measured data experiments are used to illustrate the feasibility and effectiveness of the proposed millimeter-wave multi-baseline InSAR system and the interference data processing method for large-scale mapping missions. The characteristics of airborne millimeter-wave Interferometric Synthetic Aperture Radar (InSAR) include unrestricted light, large surveying width, and high mapping precision. In recent years, with the continuous development and improvement of the technology of airborne millimeter-wave InSAR, it has gradually become a widely used mapping method. The core of the system design of a high-precision millimeter-wave InSAR system designed for small aircraft platforms comprises InSAR baseline configuration, multi-baseline configuration, the external Digital Elevation Model (DEM), and InSAR processing flow. In this study, interferometric elevation measurements influenced by different baseline parameters of an airborne millimeter-wave InSAR system are analyzed. A design scheme of the millimeter-wave multi-baseline InSAR system based on integrated antenna pod is provided. Then, a time-domain imaging algorithm-based millimeter-wave multi-baseline InSAR elevation measurement process is proposed. Finally, real measured data experiments are used to illustrate the feasibility and effectiveness of the proposed millimeter-wave multi-baseline InSAR system and the interference data processing method for large-scale mapping missions.
When the Permanent Scatterer (PS) technique is utilized to compensate the Atmospheric Phase (AP) for Ground-Based Interferometric Synthetic Aperture Radar (GB-InSAR) images, a proper parametric model should be built to describe the AP. However, for some interferograms, the AP may nonlinearly vary with the PS range, and this cannot be effectively compensated via the conventional method. This paper proposes an improved method to compensate the nonlinear AP. Here, the conventional method is first used to compensate all the phase interferograms. By calculating the standard deviation of the phase sequence of every PS and setting a proper threshold, a large number of stable PSs are selected. Then these stable PSs are divided into a certain number of sub-regions, and some control points are determined. With the inverse distance weighting interpolation, the APs of all the PSs are estimated and compensated. To verify the effectiveness of the proposed method, 460 radar images are processed, and the results are made compared with those of the conventional method. The nonlinear AP could be better compensated with the proposed method to avoid misunderstanding of the motional area. Several reference PSs are selected to make quantitative comparisons, and measurement error up to 1 rad could be reduced. When the Permanent Scatterer (PS) technique is utilized to compensate the Atmospheric Phase (AP) for Ground-Based Interferometric Synthetic Aperture Radar (GB-InSAR) images, a proper parametric model should be built to describe the AP. However, for some interferograms, the AP may nonlinearly vary with the PS range, and this cannot be effectively compensated via the conventional method. This paper proposes an improved method to compensate the nonlinear AP. Here, the conventional method is first used to compensate all the phase interferograms. By calculating the standard deviation of the phase sequence of every PS and setting a proper threshold, a large number of stable PSs are selected. Then these stable PSs are divided into a certain number of sub-regions, and some control points are determined. With the inverse distance weighting interpolation, the APs of all the PSs are estimated and compensated. To verify the effectiveness of the proposed method, 460 radar images are processed, and the results are made compared with those of the conventional method. The nonlinear AP could be better compensated with the proposed method to avoid misunderstanding of the motional area. Several reference PSs are selected to make quantitative comparisons, and measurement error up to 1 rad could be reduced.
With the development of artificial intelligence, Synthetic-Aperture Radar (SAR) ship detection using deep learning technology can effectively avoid traditionally complex feature design and thereby greatly improve detection accuracy. However, most existing detection models often improve detection accuracy at the expense of detection speed that limits some real-time applications of SAR such as emergency military deployment, rapid maritime rescue, and real-time marine environmental monitoring. To solve this problem, a high-speed and high-accuracy SAR ship detection method called SARShipNet-20 based on a Depthwise Separable Convolution Neural Network (DS-CNN) has been proposed in this paper, that replaces the Traditional Convolution Neural Network (T-CNN) and combines Channel Attention (CA) and Spatial Attention (SA). As a result, high-speed and high-accuracy SAR ship detection can be simultaneously achieved. This method has certain practical significance in the field of real-time SAR application, and its lightweight model is helpful for future FPGA or DSP hardware transplantation. With the development of artificial intelligence, Synthetic-Aperture Radar (SAR) ship detection using deep learning technology can effectively avoid traditionally complex feature design and thereby greatly improve detection accuracy. However, most existing detection models often improve detection accuracy at the expense of detection speed that limits some real-time applications of SAR such as emergency military deployment, rapid maritime rescue, and real-time marine environmental monitoring. To solve this problem, a high-speed and high-accuracy SAR ship detection method called SARShipNet-20 based on a Depthwise Separable Convolution Neural Network (DS-CNN) has been proposed in this paper, that replaces the Traditional Convolution Neural Network (T-CNN) and combines Channel Attention (CA) and Spatial Attention (SA). As a result, high-speed and high-accuracy SAR ship detection can be simultaneously achieved. This method has certain practical significance in the field of real-time SAR application, and its lightweight model is helpful for future FPGA or DSP hardware transplantation.
Over the recent years, deep-learning technology has been widely used. However, in research based on Synthetic Aperture Radar (SAR) ship target detection, it is difficult to support the training of a deep-learning network model because of the difficulty in data acquisition and the small scale of the samples. This paper provides a SAR ship detection dataset with a high resolution and large-scale images. This dataset comprises 31 images from Gaofen-3 satellite SAR images, including harbors, islands, reefs, and the sea surface in different conditions. The backgrounds include various scenarios such as the near shore and open sea. We conducted experiments using both traditional detection algorithms and deep-learning algorithms and observed the densely connected end-to-end neural network to achieve the highest average precision of 88.1%. Based on the experiments and performance analysis, corresponding benchmarks are provided as a basis for further research on SAR ship detection using this dataset. Over the recent years, deep-learning technology has been widely used. However, in research based on Synthetic Aperture Radar (SAR) ship target detection, it is difficult to support the training of a deep-learning network model because of the difficulty in data acquisition and the small scale of the samples. This paper provides a SAR ship detection dataset with a high resolution and large-scale images. This dataset comprises 31 images from Gaofen-3 satellite SAR images, including harbors, islands, reefs, and the sea surface in different conditions. The backgrounds include various scenarios such as the near shore and open sea. We conducted experiments using both traditional detection algorithms and deep-learning algorithms and observed the densely connected end-to-end neural network to achieve the highest average precision of 88.1%. Based on the experiments and performance analysis, corresponding benchmarks are provided as a basis for further research on SAR ship detection using this dataset.

微信 | 公众平台

随时查询稿件 获取最新论文 知晓行业信息