2013 Vol. 2, No. 2

Reviews
The objective of this paper is to investigate the connotation, necessity and tendency of Synthetic Aperture Radar (SAR) imaging technology within the framework of multidimensional space joint-observation, which are polarimetry space, frequency space, angle space and time space, on the basis of the key evolvement phases of SAR imaging technology. Furthermore,the definition of Multidimensional Space Joint-observation SAR (MSJosSAR) is given based on the demand of information fusion of multidimensional space joint observation SAR images. After that, the advantage of MSJosSAR is revealed by using Kronecker product decomposition technology for better understanding of target scattering mechanisms. Besides, the hypothesis and basic framework on which the MSJosSAR signal processing rely is listed and illustrated. Finally, the number of joint observation spaces of typical SAR configurations is enumerated. The objective of this paper is to investigate the connotation, necessity and tendency of Synthetic Aperture Radar (SAR) imaging technology within the framework of multidimensional space joint-observation, which are polarimetry space, frequency space, angle space and time space, on the basis of the key evolvement phases of SAR imaging technology. Furthermore,the definition of Multidimensional Space Joint-observation SAR (MSJosSAR) is given based on the demand of information fusion of multidimensional space joint observation SAR images. After that, the advantage of MSJosSAR is revealed by using Kronecker product decomposition technology for better understanding of target scattering mechanisms. Besides, the hypothesis and basic framework on which the MSJosSAR signal processing rely is listed and illustrated. Finally, the number of joint observation spaces of typical SAR configurations is enumerated.
The current status and the domestic research problems of air-borne Synthetic Aperture Ladar (SAL) are introduced and key technologies are analyzed. Based on a principle prototype, the system implementation scheme with a combination of SAR electronics technology and optical technology is given. The future practical system metrics and technological approaches are analyzed. Besides, application direction of Synthetic Aperture Ladar is clarified. The current status and the domestic research problems of air-borne Synthetic Aperture Ladar (SAL) are introduced and key technologies are analyzed. Based on a principle prototype, the system implementation scheme with a combination of SAR electronics technology and optical technology is given. The future practical system metrics and technological approaches are analyzed. Besides, application direction of Synthetic Aperture Ladar is clarified.
Special Topic Papers:Synthetic Aperture Radar (SAR)
Compared to the monostatic radar, bistatic radar has many special characteristics because of its spatial complexity. Bistatic Inverse Synthetic Aperture Radar (Bi-ISAR) can be employed as a radar imaging tool for obtaining non-cooperative target images. In this study, we first analyze the range and azimuth resolution of a Bi-ISAR system. To analyze this azimuth resolution and its spatial-variety characteristic, a definition called con-Doppler bandwidth is introduced, which helps overcome the difficulty of the targets viewing angle diversity calculation. Then, a detailed investigation is conducted to study the micro-Doppler effect caused by the vibration and the rotation of the target in the Bi-ISAR system. By comparing the difference in the micro-Doppler effect between the Bi-ISAR system and the Mono-ISAR system, we modify the extended Hough transform to extract the real micro-motion features of the targets. Finally, we provide some simulation results to validate the theoretical derivation and to illustrate the effectiveness of the proposed method. Compared to the monostatic radar, bistatic radar has many special characteristics because of its spatial complexity. Bistatic Inverse Synthetic Aperture Radar (Bi-ISAR) can be employed as a radar imaging tool for obtaining non-cooperative target images. In this study, we first analyze the range and azimuth resolution of a Bi-ISAR system. To analyze this azimuth resolution and its spatial-variety characteristic, a definition called con-Doppler bandwidth is introduced, which helps overcome the difficulty of the targets viewing angle diversity calculation. Then, a detailed investigation is conducted to study the micro-Doppler effect caused by the vibration and the rotation of the target in the Bi-ISAR system. By comparing the difference in the micro-Doppler effect between the Bi-ISAR system and the Mono-ISAR system, we modify the extended Hough transform to extract the real micro-motion features of the targets. Finally, we provide some simulation results to validate the theoretical derivation and to illustrate the effectiveness of the proposed method.
Combined with Frequency-Modulated Continuous-Wave (FMCW) technology, airborne forward-looking Synthetic Aperture Radar (SAR) with linear array antennas can obtain the image in front of the aircraft and also have the advantages of FMCW radar such as small size and lightweight. Moreover, it is suitable to be installed on platform like helicopter and small unmanned aerial vehicle. Motion compensation for forward-looking SAR with linear array antennas is one of the key problems to obtain the image in front of the aircraft in practice. This paper analyses the influence of motion error in aircraft on echo model based on the geometry of forward looking SAR with linear array antennas, and proposes a motion compensation scheme. Moreover, the compensation scheme is applicable to an improved frequency scaling algorithm (FSA) for FMCW forward looking SAR with linear array antennas. Finally, the compensation scheme is verified with the simulation. Combined with Frequency-Modulated Continuous-Wave (FMCW) technology, airborne forward-looking Synthetic Aperture Radar (SAR) with linear array antennas can obtain the image in front of the aircraft and also have the advantages of FMCW radar such as small size and lightweight. Moreover, it is suitable to be installed on platform like helicopter and small unmanned aerial vehicle. Motion compensation for forward-looking SAR with linear array antennas is one of the key problems to obtain the image in front of the aircraft in practice. This paper analyses the influence of motion error in aircraft on echo model based on the geometry of forward looking SAR with linear array antennas, and proposes a motion compensation scheme. Moreover, the compensation scheme is applicable to an improved frequency scaling algorithm (FSA) for FMCW forward looking SAR with linear array antennas. Finally, the compensation scheme is verified with the simulation.
motion errors, frequently occur in SAR images. For airborne SAR system with very high resolution and airborne repeat-pass SAR interferometry, the residual motion errors must be estimated and compensated. Multi-squint Technique with Point Targets (MTPT) is able to estimate the residual motion errors for an individual SAR image, but errors in the platform velocity and the slant range will deteriorate the accuracy of the method. Based on the detailed analysis of the velocity and slant range to residual motion error estimation, simulated and real SAR data are used to validate it in this paper. It is also pointed out that though MTPT is able to estimate the errors in the velocity and slant range, it is sensitive to the phase error. Therefore, it is advised that the errors in the velocity and slant range should be removed with other precise method before MTPT is utilized to estimate the residual motion errors. motion errors, frequently occur in SAR images. For airborne SAR system with very high resolution and airborne repeat-pass SAR interferometry, the residual motion errors must be estimated and compensated. Multi-squint Technique with Point Targets (MTPT) is able to estimate the residual motion errors for an individual SAR image, but errors in the platform velocity and the slant range will deteriorate the accuracy of the method. Based on the detailed analysis of the velocity and slant range to residual motion error estimation, simulated and real SAR data are used to validate it in this paper. It is also pointed out that though MTPT is able to estimate the errors in the velocity and slant range, it is sensitive to the phase error. Therefore, it is advised that the errors in the velocity and slant range should be removed with other precise method before MTPT is utilized to estimate the residual motion errors.
In this paper, we present the methods of parametrically reconstructing the position of a 3D scattering center model of the targets from SAR images for a target recognition application. First, a framework for reconstructing the 3D position of the targets from the SAR images is proposed. Then, a method that uses the parameters obtained from the scattering center of the targets from the SAR images is used for reconstructing the 3D scattering center model. Finally, the proposed methods are confirmed using the simulated data. In this paper, we present the methods of parametrically reconstructing the position of a 3D scattering center model of the targets from SAR images for a target recognition application. First, a framework for reconstructing the 3D position of the targets from the SAR images is proposed. Then, a method that uses the parameters obtained from the scattering center of the targets from the SAR images is used for reconstructing the 3D scattering center model. Finally, the proposed methods are confirmed using the simulated data.
Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) image is investigated. A SAR feature extraction algorithm based on multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network, Restricted Boltzmann Machine (RBM) modeling probability distribution of environment. Through the formation of more expressive multilayer neural network, the deep learning model learns shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of proposed algorithm. Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) image is investigated. A SAR feature extraction algorithm based on multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network, Restricted Boltzmann Machine (RBM) modeling probability distribution of environment. Through the formation of more expressive multilayer neural network, the deep learning model learns shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of proposed algorithm.
The circular-scanning Synthetic Aperture Radar (SAR) is a novel working mode and its image quality is closely related to the accuracy of the imaging parameters, especially considering the inaccuracy of the real speed of the motion. According to the characteristics of the circular-scanning mode, a new data based method for estimating the velocities of the radar platform and the scanning-angle of the radar antenna is proposed in this paper. By referring to the basic conception of the Doppler navigation technique, the mathematic model and formulations for the parameter estimation are firstly improved. The optimal parameter approximation based on the least square criterion is then realized in solving those equations derived from the data processing. The simulation results verified the validity of the proposed scheme. The circular-scanning Synthetic Aperture Radar (SAR) is a novel working mode and its image quality is closely related to the accuracy of the imaging parameters, especially considering the inaccuracy of the real speed of the motion. According to the characteristics of the circular-scanning mode, a new data based method for estimating the velocities of the radar platform and the scanning-angle of the radar antenna is proposed in this paper. By referring to the basic conception of the Doppler navigation technique, the mathematic model and formulations for the parameter estimation are firstly improved. The optimal parameter approximation based on the least square criterion is then realized in solving those equations derived from the data processing. The simulation results verified the validity of the proposed scheme.
Data processing is a time-consuming matter in the field of Synthetic Aperture Radar (SAR). In other ways, Graphic Processing Unit (GPU) have tremendous float-point computational horsepower and very high memory bandwidth, and the developing Compute Unified Device Architecture (CUDA) technology has enabled the GPU to be applied to the general purpose parallel computing. A new method of processing SAR data on GPU is presented in this paper. Compared with the nominal GPU based SAR processing method, number of data transfers between CPU/GPU are reduced from 4 to 1, and CPUs are exploited to cooperate with GPU synchronously. By the proposed method, data Processing is speeded up by 2.3 times, which is verified by the testing on the simulated SAR data. Data processing is a time-consuming matter in the field of Synthetic Aperture Radar (SAR). In other ways, Graphic Processing Unit (GPU) have tremendous float-point computational horsepower and very high memory bandwidth, and the developing Compute Unified Device Architecture (CUDA) technology has enabled the GPU to be applied to the general purpose parallel computing. A new method of processing SAR data on GPU is presented in this paper. Compared with the nominal GPU based SAR processing method, number of data transfers between CPU/GPU are reduced from 4 to 1, and CPUs are exploited to cooperate with GPU synchronously. By the proposed method, data Processing is speeded up by 2.3 times, which is verified by the testing on the simulated SAR data.
Compressive sensing SAR imaging can significantly reduce the sampling rate and the amount of data,but it is essential only in the case where the reflection coefficients of SAR scene are sparse. This paper proposed a compressive sensing SAR imaging method based on wavelet packet sparse representation. The wavelet packet algorithm is used to choose the most sparse representation of the SAR scene by training the same type of SAR images. By solving for the minimum 1 l norm optimization, the SAR scene reflection coefficients can be reconstructed. Unambiguous SAR image can be produced with the proposed method even with fewer samples. SAR data simulation experiments demonstrate the efficiency of the proposed method. Compressive sensing SAR imaging can significantly reduce the sampling rate and the amount of data,but it is essential only in the case where the reflection coefficients of SAR scene are sparse. This paper proposed a compressive sensing SAR imaging method based on wavelet packet sparse representation. The wavelet packet algorithm is used to choose the most sparse representation of the SAR scene by training the same type of SAR images. By solving for the minimum 1 l norm optimization, the SAR scene reflection coefficients can be reconstructed. Unambiguous SAR image can be produced with the proposed method even with fewer samples. SAR data simulation experiments demonstrate the efficiency of the proposed method.
The Synthetic Aperture Radar (SAR) provides abundant data for the research of eddies. It is important to extract the information of eddies in the SAR image effectively. In this paper, a method based on logarithmic spiral edge fitting for information extraction of eddy in the SAR image is presented, which is capable to extract the eddy center, eddy diameter, and the size of eddy edge. Based on this method, an experiment is conducted to extract the information of eddy in the sequential SAR images acquired by ENVISAT ASAR and ERS-2. The information of two eddies and their movement can be achieved. The validity of the method can be verified by comparing with the pseudo-color composite results. The Synthetic Aperture Radar (SAR) provides abundant data for the research of eddies. It is important to extract the information of eddies in the SAR image effectively. In this paper, a method based on logarithmic spiral edge fitting for information extraction of eddy in the SAR image is presented, which is capable to extract the eddy center, eddy diameter, and the size of eddy edge. Based on this method, an experiment is conducted to extract the information of eddy in the sequential SAR images acquired by ENVISAT ASAR and ERS-2. The information of two eddies and their movement can be achieved. The validity of the method can be verified by comparing with the pseudo-color composite results.
Paper
Collision avoidance radar for trains is pregnant for safety transportation. In order to realize low cost and high performance of azimuth accuracy, we have developed MMW (Milli-Meter Wave) radar, which employs switched phased array and frequency stepped technology. This paper analyses the radiation patterns of transmitting/receiving antennas and compensation method for amplitude/phase errors of synthetic wideband frequency stepped signal. To confirm the operation of the radar, low cost millimeter-wave collision avoidance radar was fabricated. Lots of experiments confirmed a high azimuth and range resolution. Collision avoidance radar for trains is pregnant for safety transportation. In order to realize low cost and high performance of azimuth accuracy, we have developed MMW (Milli-Meter Wave) radar, which employs switched phased array and frequency stepped technology. This paper analyses the radiation patterns of transmitting/receiving antennas and compensation method for amplitude/phase errors of synthetic wideband frequency stepped signal. To confirm the operation of the radar, low cost millimeter-wave collision avoidance radar was fabricated. Lots of experiments confirmed a high azimuth and range resolution.
The best-fitting Gaussian approximation Probability Hypothesis Density (PHD) filter is a novel algorithm for multiple maneuvering target tracking. However, there is a problem that the model probabilities are calculated without the measurement innovation. To solve this problem, an improved algorithm is proposed in this paper, which develops an update procedure for model probabilities to employ the posterior measurement innovation to enhance the filtering performance. Then, the dynamic equations can be softly switched among different models according to the likelihood functions. The simulation results show that the improved algorithm has the advantages over the ordinary one in the aspects of target number estimation and filtering accuracy, implying good application prospect. The best-fitting Gaussian approximation Probability Hypothesis Density (PHD) filter is a novel algorithm for multiple maneuvering target tracking. However, there is a problem that the model probabilities are calculated without the measurement innovation. To solve this problem, an improved algorithm is proposed in this paper, which develops an update procedure for model probabilities to employ the posterior measurement innovation to enhance the filtering performance. Then, the dynamic equations can be softly switched among different models according to the likelihood functions. The simulation results show that the improved algorithm has the advantages over the ordinary one in the aspects of target number estimation and filtering accuracy, implying good application prospect.
A novel method of moving targets detection taking Doppler spectrum analysis technique for Passive Coherent Radar (PCR) is provided. After dividing the receiving signals into segments as pulse series, it utilizes the technique of pulse compress and Doppler processing to detect and locate the targets. Based on the algorithm for Pulse-Doppler (PD) radar, the equipollence between continuous and pulsed wave in match filtering is proved and details of this method are introduced. To compare it with the traditional method of Cross-Ambiguity Function (CAF) calculation, the relationship and mathematical modes of them are analyzed, with some suggestions on parameters choosing. With little influence to the gain of targets, the method can greatly promote the processing efficiency. The validity of the proposed method is demonstrated by offline processing real collected data sets and simulation results. A novel method of moving targets detection taking Doppler spectrum analysis technique for Passive Coherent Radar (PCR) is provided. After dividing the receiving signals into segments as pulse series, it utilizes the technique of pulse compress and Doppler processing to detect and locate the targets. Based on the algorithm for Pulse-Doppler (PD) radar, the equipollence between continuous and pulsed wave in match filtering is proved and details of this method are introduced. To compare it with the traditional method of Cross-Ambiguity Function (CAF) calculation, the relationship and mathematical modes of them are analyzed, with some suggestions on parameters choosing. With little influence to the gain of targets, the method can greatly promote the processing efficiency. The validity of the proposed method is demonstrated by offline processing real collected data sets and simulation results.
The approach of tracking maneuvering targets based on the Current Statistical (CS) model is widely used. The method needs to preset maneuvering frequency and maximum acceleration based on experience. Inpractice, the preset values are often not consistent with the actual moving state of targets and result in larger tracking errors. In order to tackle the problem, this paper initially deduces a self-adapting maneuvering frequency algorithm from the discrete state equation of the CS model. Then, an improved self-adapting acceleration covariance algorithm is presented. Simulation results show that, by using the self-adapting maneuvering frequency algorithm and the improved self-adapting acceleration covariance algorithm to track targets simultaneously, the ability to self-adapt to the fluctuation of the moving state will be improved. The tracking accuracy is also improved, and the convergence speed of the algorithm is quicker. The approach of tracking maneuvering targets based on the Current Statistical (CS) model is widely used. The method needs to preset maneuvering frequency and maximum acceleration based on experience. Inpractice, the preset values are often not consistent with the actual moving state of targets and result in larger tracking errors. In order to tackle the problem, this paper initially deduces a self-adapting maneuvering frequency algorithm from the discrete state equation of the CS model. Then, an improved self-adapting acceleration covariance algorithm is presented. Simulation results show that, by using the self-adapting maneuvering frequency algorithm and the improved self-adapting acceleration covariance algorithm to track targets simultaneously, the ability to self-adapt to the fluctuation of the moving state will be improved. The tracking accuracy is also improved, and the convergence speed of the algorithm is quicker.