Deep convolutional neural network for meteorology target detection in airborne weather radar images

Considering the problem that the scattering echo images of airborne Doppler weather radar are often reduced by ground clutters,the accuracy and confidence of meteorology target detection are reduced.In this paper,a deep convolutional neural network(DCNN)is proposed for meteorology target detection and ground clutter suppression with a large collection of airborne weather radar images as network input.For each weather radar image,the corresponding digital elevation model(DEM)image is extracted on basis of the radar antenna scan-ning parameters and plane position,and is further fed to the net-work as a supplement for ground clutter suppression.The fea-tures of actual meteorology targets are learned in each bottle-neck module of the proposed network and convolved into deeper iterations in the forward propagation process.Then the network parameters are updated by the back propagation itera-tion of the training error.Experimental results on the real mea-sured images show that our proposed DCNN outperforms the counterparts in terms of six evaluation factors.Meanwhile,the network outputs are in good agreement with the expected mete-orology detection results(labels).It is demonstrated that the pro-posed network would have a promising meteorology observa-tion application with minimal effort on network variables or parameter changes.

Lira-YOLO: a lightweight model for ship detection in radar images

For the detection of marine ship objects in radar images, large-scale networks based on deep learning are difficult to be deployed on existing radar-equipped devices. This paper proposes a lightweight convolutional neural network, LiraNet, which combines the idea of dense connections, residual connections and group convolution, including stem blocks and extractor modules.The designed stem block uses a series of small convolutions to extract the input image features, and the extractor network adopts the designed two-way dense connection module, which further reduces the network operation complexity. Mounting LiraNet on the object detection framework Darknet, this paper proposes Lira-you only look once(Lira-YOLO), a lightweight model for ship detection in radar images, which can easily be deployed on the mobile devices. Lira-YOLO's prediction module uses a two-layer YOLO prediction layer and adds a residual module for better feature delivery. At the same time, in order to fully verify the performance of the model, mini-RD, a lightweight distance Doppler domain radar images dataset, is constructed. Experiments show that the network complexity of Lira-YOLO is low, being only 2.980 Bflops, and the parameter quantity is smaller, which is only 4.3 MB. The mean average precision(mAP) indicators on the mini-RD and SAR ship detection dataset(SSDD) reach 83.21% and 85.46%, respectively,which is comparable to the tiny-YOLOv3. Lira-YOLO has achieved a good detection accuracy with less memory and computational cost.

Investigation of Simulating Radar Images Concerning the Multipath Scattering Effect

In the composed system of a target and rough surface, the electromagnetic scattering mechanism, especially the multipath scattering, is investigated. Using physical optics double bouncing algorithm, the multipath scattering model of the system has been established. Simulated by a wide-band radar signal and based on fractal rough surface, the artificial echo of the target has been obtained in virtue of the established multipath scattering model. By simulating to image the target in one dimension using the artificial echo, two kinds of range profiles are attained. It is found that one is from the target and the other is from the multipath scattering effect. Key words multipath scattering - radar imaging - rough surface scattering CLC number O 451 Foundation item: Supported by the Key Laboratory Foundation of National Defense Science and Technology (99JS93. 1. 2. JW1204)Biography: Yang Chun-hua(1978-), male, Ph. D candidate, research direction: radio wave propagation and wiresless communication.

A novel algorithm for ocean wave direction inversion from X-band radar images based on optical flow method

As one of the important sea state parameters for navigation safety and coastal resource management, the ocean wave direction represents the propagation direction of the wave. A novel algorithm based on an optical flow method is developed for the ocean wave direction inversion of the ocean wave fields imaged by the X-band radar continuously. The proposed algorithm utilizes the echo images received by the X-band wave monitoring radar to estimate the optical flow motion, and then the actual wave propagation direction can be obtained by taking a weighted average of the motion vector for each pixel. Compared with the traditional ocean wave direction inversion method based on frequency-domain, the novel algorithm is fully using a time-domain signal processing method without determination of a current velocity and a modulation transfer function（MTF）. In the meantime,the novel algorithm is simple, efficient and there is no need to do something more complicated here. Compared with traditional ocean wave direction inversion method, the ocean wave direction of derived by using this proposed method matches well with that measured by an in situ buoy nearby and the simulation data. These promising results demonstrate the efficiency and accuracy of the algorithm proposed in the paper.

Sea surface reconstruction from marine radar images using deep convolutional neural networks

The sea surface reconstructed from radar images provides valuable information for marine operations and maritime transport.The standard reconstruction method relies on the three-dimensional fast Fourier transform(3D-FFT),which introduces empirical parameters and modulation transfer function(MTF)to correct the modulation effects that may cause errors.In light of the convolutional neural networks’(CNN)success in computer vision tasks,this paper proposes a novel sea surface reconstruction method from marine radar images based on an end-to-end CNN model with the U-Net architecture.Synthetic radar images and sea surface elevation maps were used for training and testing.Compared to the standard reconstruction method,the CNN-based model achieved higher accuracy on the same data set,with an improved correlation coefficient between reconstructed and actual wave fields of up to 0.96-0.97,and a decreased non-dimensional root mean square error(NDRMSE)of around 0.06.The influence of training data on the deep learning model was also studied.Additionally,the impact of the significant wave height and peak period on the CNN model’s accuracy was investigated.It has been demonstrated that the accuracy will fluctuate as the wave steepness increases,but the correlation coefficient remains above 0.90,and the NDRMSE remains less than 0.11.

Three-dimensional positions of scattering centers reconstruction from multiple SAR images based on radargrammetry

A method and procedure is presented to reconstruct three-dimensional(3D) positions of scattering centers from multiple synthetic aperture radar(SAR) images. Firstly, two-dimensional(2D) attribute scattering centers of targets are extracted from 2D SAR images. Secondly, similarity measure is developed based on 2D attributed scatter centers' location, type, and radargrammetry principle between multiple SAR images. By this similarity, we can associate 2D scatter centers and then obtain candidate 3D scattering centers. Thirdly, these candidate scattering centers are clustered in 3D space to reconstruct final 3D positions. Compared with presented methods, the proposed method has a capability of describing distributed scattering center, reduces false and missing 3D scattering centers, and has fewer restrictionson modeling data. Finally, results of experiments have demonstrated the effectiveness of the proposed method.

A Heterogeneous Information Fusion Method for Maritime Radar and AIS Based on D-S Evidence Theory

Maritime radar and automatic identification systems (AIS), which are essential auxiliary equipment for navigation safety in the shipping industry, have played significant roles in maritime safety supervision. However, in practical applications, the information obtained by a single device is limited, and it is necessary to integrate the information of maritime radar and AIS messages to achieve better recognition effects. In this study, the D-S evidence theory is used to fusion the two kinds of heterogeneous information: maritime radar images and AIS messages. Firstly, the radar image and AIS message are processed to get the targets of interest in the same coordinate system. Then, the coordinate position and heading of targets are chosen as the indicators for judging target similarity. Finally, a piece of D-S evidence theory based on the information fusion method is proposed to match the radar target and the AIS target of the same ship. Particularly, the effectiveness of the proposed method has been validated and evaluated through several experiments, which proves that such a method is practical in maritime safety supervision.

Electromagnetic scattering and imaging simulation of extremely large-scale sea-ship scene based on GPU parallel technology

Aiming to solve the bottleneck problem of electromagnetic scattering simulation in the scenes of extremely large-scale seas and ships,a high-frequency method by using graphics processing unit(GPU)parallel acceleration technique is proposed.For the implementation of different electromagnetic methods of physical optics(PO),shooting and bouncing ray(SBR),and physical theory of diffraction(PTD),a parallel computing scheme based on the CPU-GPU parallel computing scheme is realized to balance computing tasks.Finally,a multi-GPU framework is further proposed to solve the computational difficulty caused by the massive number of ray tubes in the ray tracing process.By using the established simulation platform,signals of ships at different seas are simulated and their images are achieved as well.It is shown that the higher sea states degrade the averaged peak signal-to-noise ratio(PSNR)of radar image.

Power of SAR Imagery and Machine Learning in Monitoring Ulva prolifera:A Case Study of Sentinel-1 and Random Forest

Automatically detecting Ulva prolifera(U.prolifera)in rainy and cloudy weather using remote sensing imagery has been a long-standing problem.Here,we address this challenge by combining high-resolution Synthetic Aperture Radar(SAR)imagery with the machine learning,and detect the U.prolifera of the South Yellow Sea of China(SYS)in 2021.The findings indicate that the Random Forest model can accurately and robustly detect U.prolifera,even in the presence of complex ocean backgrounds and speckle noise.Visual inspection confirmed that the method successfully identified the majority of pixels containing U.prolifera without misidentify-ing noise pixels or seawater pixels as U.prolifera.Additionally,the method demonstrated consistent performance across different im-ages,with an average Area Under Curve(AUC)of 0.930(+0.028).The analysis yielded an overall accuracy of over 96%,with an aver-age Kappa coefficient of 0.941(+0.038).Compared to the traditional thresholding method,Random Forest model has a lower estima-tion error of 14.81%.Practical application indicates that this method can be used in the detection of unprecedented U.prolifera in 2021 to derive continuous spatiotemporal changes.This study provides a potential new method to detect U.prolifera and enhances our under-standing of macroalgal outbreaks in the marine environment.

Entropy function optimization for radar imaging

The convergence performance of the minimum entropy auto-focusing（MEA） algorithm for inverse synthetic aperture radar（ISAR） imaging is analyzed by simulation. The results show that a local optimal solution problem exists in the MEA algorithm. The cost function of the MEA algorithm is not a downward-convex function of multidimensional phases to be compensated. Only when the initial values of the compensated phases are chosen to be near the global minimal point of the entropy function, the MEA algorithm can converge to a global optimal solution. To study the optimal solution problem of the MEA algorithm, a new scheme of entropy function optimization for radar imaging is presented. First, the initial values of the compensated phases are estimated by using the modified Doppler centroid tracking （DCT）algorithm. Since these values are obtained according to the maximum likelihood （ML） principle, the initial phases can be located near the optimal solution values. Then, a fast MEA algorithm is used for the local searching process and the global optimal solution can be obtained. The simulation results show that this scheme can realize the global optimization of the MEA algorithm and can avoid the selection and adjustment of parameters such as iteration step lengths, threshold values, etc.

Recognition of the major scattering sources on complex targets based on the high frequency radar cross section integrated calculation technique

Based on the high frequency （HF） integrated radar cross section （RCS） calculation approach, a technique of detecting major scattering source is developed by using an appropriate arithmetic for scattering distribution and scattering source detection. For the perfect adaptability to targets and the HF of the HF integrated RCS calculation platform, this technique is suitable to solve large complex targets and has lower requirement to the target modeling. A comparison with the result of 2-D radar imaging confirms the accuracy and reliability of this technique in recognition of the major scattering source on complex targets. This technique provides the foundation for rapid integrated evaluation of the scattering performance and 3-D scattering model reconstruction of large complex targets.

Application of Superresolution Techniques to Radar Imaging

Modern spectral estimation techniques (superresolution in technical jargon) have been applied to many fields of signal processing since many years[1][2]. Application to radar imaging, mainlyto ISAR (Inverse Synthetic Aperture Radar) is documented in some recent papers[3] to [6]. Applications have been attempted also to SAR (Synthetic Aperture Radar)[7][8]. In these fields the benefit ofspectral estimation reveals in a resolution beyond the Rayleigh limits set by compressed pulse andsynthetic aperture lengths. Furthermore very low sidelobes of point scatterer response are obtained.In this paper superresolution has been applied both to simulated stepped-frequency ISAR dataand to real ERS-1 SAR data; the achieved results are encouraging and suggest a more extensivepractical application of the technique. The paper is organized in two parts. In the first we have applied the autoregressive (AR) and the minimum variance (MV)-Capon methods to improve therange resolution of simulated ISAR data. In the second part we have conceived an upgraded versionof spectral analysis (SPECAN) processing to obtain a SAR image of better quality. The method hasbeen tested on recorded live data of ERS-1 mission.

Novel imaging methods of stepped frequency radar based on compressed sensing

The theory of compressed sensing （CS） provides a new chance to reduce the data acquisition time and improve the data usage factor of the stepped frequency radar system. In light of the sparsity of radar target reflectivity, two imaging methods based on CS, termed the CS-based 2D joint imaging algorithm and the CS-based 2D decoupled imaging algorithm, are proposed. These methods incorporate the coherent mixing operation into the sparse dictionary, and take random measurements in both range and azimuth directions to get high resolution radar images, thus can remarkably reduce the data rate and simplify the hardware design of the radar system while maintaining imaging quality. Ex- periments from both simulated data and measured data in the anechoic chamber show that the proposed imaging methods can get more focused images than the traditional fast Fourier trans- form method. Wherein the joint algorithm has stronger robustness and can provide clearer inverse synthetic aperture radar images, while the decoupled algorithm is computationally more efficient but has slightly degraded imaging quality, which can be improved by increasing measurements or using a robuster recovery algorithm nevertheless.

Micro-motion effect in inverse synthetic aperture radar imaging of ballistic mid-course targets

For ballistic mid-course targets,in addition to constant orbital motion,the target or any structure on the target undergoes micro-motion dynamics,such as spin,precession and tumbling.The micro-motion characteristics of the ballistic mid-course targets were discussed.The target motion model and inverse synthetic aperture radar(ISAR) imaging model for this kind of targets were built.Then,the influence of micro-motion on ISAR imaging based on the established imaging model was presented.The computer simulation to get mid-course target echoes from static darkroom electromagnetic scattering data based on the established target motion model was realized.The imaging results of computer simulation show the validity of ISAR imaging analysis for micro-motion targets.

Parameter estimation for rigid body after micro-Doppler removal based on L-statistics in the radar analysis

In traditional inverse synthetic aperture radar （ISAR） imaging of moving targets with rotational parts, the micro-Doppler （m-D） effects caused by the rotational parts influence the quality of the radar images. Recently, L. Stankovic proposed an m-D removal method based on L-statistics, which has been proved effective and simple. The algorithm can extract the m-D effects according to different behaviors of signals induced by rotational parts and rigid bodies in time-frequency （T-F） domain. However, by removing m-D effects, some useful short time Fourier transform （STFT） samples of rigid bodies are also extracted, which induces the side lobe problem of rigid bodies. A parameter estimation method for rigid bodies after m-D removal is proposed, which can accurately re- cover rigid bodies and avoid the side lobe problem by only using m-D removal. Simulations are given to validate the effectiveness of the proposed method.

Improved HHT and its application in narrowband radar imaging for precession cone-shaped targets

Imaging the spatial precession cone-shaped targets with narrowband radar is a new technical approach in mid-course recognition problem. However, most existing time-frequency methods still have some inevitable deficiencies for extracting microDoppler information in practical applications, which leads to blurring of the image. A new narrowband radar imaging algorithm for the precession cone-shaped targets is proposed. The instantaneous frequency of each scattering point is gained by using the improved Hilbert-Huang transform, then the positions of scattering points in the parameter domain are reconstructed. Numerical simulation and experiment results confirm the effectiveness and high precision of the proposed algorithm.

Design of synthetic aperture radar low-intercept radio frequency stealth

Not confined to a certain point,such as waveform,this paper systematically studies the low-intercept radio frequency(RF)stealth design of synthetic aperture radar(SAR)from the system level.The study is carried out from two levels.In the first level,the maximum low-intercept range equation of the conventional SAR system is deduced firstly,and then the maximum low-intercept range equation of the multiple-input multiple-output SAR system is deduced.In the second level,the waveform design and imaging method of the low-intercept RF SAR system are given and verified by simulation.Finally,the main technical characteristics of the lowintercept RF stealth SAR system are given to guide the design of low-intercept RF stealth SAR system.

MIMO RADAR IMAGING MODEL AND ALGORITHM

Based on the array architecture of multiple transmitting/receiving antennas, Multi-Input Multi-Output (MIMO) radar provides a new mechanism for radar imaging technology. In order to explore the processing approach to this imaging mechanism, the two dimensional (2D) imaging model of MIMO radar is established first, and the spatial sampling ability is analyzed from the concept of spatial convolution of the antenna elements. The target spatial spectral filling format of MIMO radar with monochromatic transmitting signal is described. High-resolution imaging capability of MIMO radar is analyzed according to spatial spectral coverage and the corresponding imaging algorithm is presented. Finally, field imaging experiment is used to demonstrate the superior imaging performance of MIMO radar.

Aliasing-free high resolution imaging of fast rotating targets with narrowband radar

Narrowband radar has been successfully used for high resolution imaging of fast rotating targets by exploiting their micro-motion features.In some practical situations,however,the target image may suffer from aliasing due to the fixed pulse repetition interval(PRI)of traditional radar scheme.In this work,the random PRI signal associated with compressed sensing(CS)theory was introduced for aliasing reduction to obtain high resolution images of fast rotating targets.To circumvent the large-scale dictionary and high computational complexity problem arising from direct application of CS theory,the low resolution image was firstly generated by applying a modified generalized Radon transform on the time-frequency domain,and then the dictionary was scaled down by random undersampling as well as the atoms extraction according to those strong scattering areas of the low resolution image.The scale-down-dictionary CS(SDD-CS)processing scheme was detailed and simulation results show that the SDD-CS scheme for narrowband radar can achieve preferable images with no aliasing as well as acceptable computational cost.

ADAPTIVE RIVER SEGMENTATION IN SAR IMAGES

There is difficulty for distinguishing of river and shadow in Synthetic Aperture Radar (SAR) images. A method of river segmentation in SAR images based on wavelet energy and gradient is proposed in this paper. It mainly includes two algorithms: coarse segmentation and refined segmen- tation. Firstly, The river regions are coarsely segmented by the wavelet energy feature,and then refined segmented accurately by the gradient threshold which is got adaptively. The experimental results show the validity of the method, which provides a good foundation for targets detection above the river.