Maosheng Xiang

Accelerating InSAR raw data simulation on GPU using CUDA

This paper describes a scalable parallel method for interferometric synthetic aperture radar (InSAR) raw data simulation on graphic processing unit (GPU) with common unified device architecture (CUDA). The advantages of the new method rely on the three contributions: GPU hardware provides lots of stream processors for threads calculating, CUDA software environment runs thousands of threads working in parallel for assigned task, raw data simulation adopts the fine-grained task parallelism. Compared with OpenMP, MPI and grid computing, the method not only improves the computational efficiency greatly, but also save the resources such as hardware, electric power and room space. The results show that the method not only ensures accuracy, but also be able to obtain the speedup about 30 times.

Hybrid general-purpose computation on GPU (GPGPU) and computer graphics synthetic aperture radar simulation for complex scenes

In this paper, a new hybrid general-purpose computation on GPU (GPGPU) and computer graphics synthetic aperture radar (SAR) simulation method for complex scenes is proposed.xa0Previous SAR simulations for complex scenes only use GPU’s graphics capabilities for scattering calculation in graphical electromagnetic computing (GRECO) algorithm. The new hybrid method use GPU’s graphics and parallel computing capabilities for geometry modeling, scattering map and raw data calculation in SAR simulation of complex scenes. The advantages of the new method rely on the three contributions: GPU hardware provides lots of stream processors for threads calculating, common unified device architecturexa0(CUDA)xa0environment runs thousands of threads working in parallel for assigned task, raw data simulation adopts the fine-grained task parallelism. Compared withxa0classical algorithms, the method not onlyxa0ensures the accuracy of scattering calculation with GRECO algorithm, but alsoxa0improves the computational efficiency greatlyfor complex scenes consideration.xa0The results show that the methodxa0isxa0able to obtain the speedupxa0about 30 timesxa0on entry-level GPU. n n xa0 n n Key words:xa0Synthetic aperture radar (SAR), raw data simulation, parallel processing, general-purpose computation on GPU (GPGPU), computer graphics.

Error analysis of SAR motion compensation

During Synthetic Aperture Radar (SAR) motion compensation processing, the residual uncompensated motion errors may degrade SAR images, and these errors mainly result from the positioning inaccuracies of the track and target. By analyzing SAR motion compensation method and the effect of the track and target positioning inaccuracies, this paper presents a model to calculate the residual uncompensated motion errors. According to the model, the residual errors mainly depend on the slant range error, Doppler centroid error, and the target height error. The computed results by the proposed model accord closely with the actual image quality.

A weighted calibration method of interferometric SAR data

The accuracy of the digital elevation model (DEM) generated by the interferometric synthetic aperture radar (SAR) partly depends on the accuracy of system parameters, so it is necessary to calibrate the system parameters. The traditional calibration method models the elevation error as a linear function of parameter biases, and solves the biases through the sensitivity equations. This paper presents a weighted calibration method applicable to interferometric SAR data. It introduces the weightings to the sensitivity equations to discriminate the ground control points (GCPs) with different correlation coefficients and locations. This weighted calibration method can improve the DEM accuracy, and this paper illustrates its successful application to airborne interferometric SAR data collected by Institute of Electronics, Chinese Academy of Sciences.

Joint interferometric calibration based on block adjustment for an airborne dual-antenna InSAR system

Mapping large areas using airborne dual-antenna interferometric synthetic aperture radar (InSAR) usually requires processing and mosaicking of different scenes from multiple strips. The overlapping areas of these multiple strips should have consistent elevation values. Due to the unstable attitude of the plane, the interferometric parameters usually vary for each scene during mapping. Therefore, interferometric calibration technology for high-precision height retrieval is required for the correction of the interferometric errors. The traditional interferometric calibration methods for a single scene usually use ground control points (GCPs) to estimate the interferometric parameters – this method cannot guarantee a consistent height in the area of overlap. Besides, GCPs are difficult to deploy over rough terrain, making it impossible to use traditional calibration methods. In this article, a joint interferometric calibration method based on the block adjustment theory used in photogrammetry is proposed for airborne dual-antenna InSAR. This method considers the accurate digital elevation model (DEM) height reconstruction model and can be applied with sparse GCPs. The principle of the proposed method is to make the best use of the GCPs within all the scenes and the tie points (TPs) between the adjacent scenes to establish an error relationship model. First, the weighting values of all GCPs and TPs based on their retrieval elevation error caused by the interferometric phase error and the position distribution difference are introduced in the proposed method. Next, the interferometric parameters are weighted to reduce the condition number of the normal equation. Then, an alternative approximation approach combined with the sparse matrix decomposition technique LDLT is utilized to solve the normal equation, and the corrected interferometric parameters for each scene are obtained. High-precision joint interferometric calibration results for airborne InSAR systems are achieved by the proposed method and validated by experiment. Using the proposed method, the average mean error (AME) and root mean square error (RMSE) are below 0.6037 and 0.9176 m, respectively. Meanwhile, the maximum AME and RMSE of the reconstructed DEM height difference for the validation TPs in the overlapped area of the adjacent scenes are reduced from 1.2909 and 1.7245 m to 0.8864 and 1.2087 m, respectively.

The mathematic model of multipath error in airborne interferometric SAR system

In airborne two-antenna interferometric SAR system, the returned pulse may be reflected by parts of the aircraft platform into the antennas, and this is called multipath effect which will cause oscillating phase errors and hence height errors. This paper presents a theoretical model to compute the multipath error. In this model the multipath phase error is a function of look angle or ideal phase, and the unknown parameters of the model can be estimated from distributed targets with known elevation. On the basis of the model, a method and processing procedure can be used to correct multipath error effectively, and this paper illustrates its successful application to interferometric SAR data collected by Institute of Electronics, Chinese Academy of Sciences.

Processing for airborne interferometric SAR data with high squint

A novel approach for the highly squinted airborne InSAR data processing is presented. Using the IMU data to resolve the PRF ambiguity and moving azimuth windows according to the Doppler centroid varying in the different range, as well as combining the auto-registration imaging algorithm, such approach not only can compensate the squint effect and motion error directly at the imaging processing stage, but also can improve the coherence and restrain the interferometric phase error of the image-pair. The simulative and practical results indicate that the proposed approach is very suitable for the processing of the data with a high squint for a dual-antenna airborne InSAR system with its efficiency in improving the image quality and enhancing the interferogram and coherence.

Effects of Target Positioning Error on Motion Compensation for Airborne Interferometric SAR: Effects of Target Positioning Error on Motion Compensation for Airborne Interferometric SAR

The measurement inaccuracies of Inertial Measurement Unit/Global Positioning System (IMU/GPS) as well as the positioning error of the target may contribute to the residual uncompensated motion errors in the MOtion COmpensation (MOCO) approach. Aiming at the effects of target positioning error on MOCO for airborne interferometric SAR, the paper deduces a mathematical model of the residual motion error caused by the target positioning error under the condition of squint. Also, the effects of the system sampling delay, Doppler center frequency, and reference DEM errors on the residual motion error that result in target positioning error based on the model are analyzed. Then, the paper discusses the effects of the reference DEM error on the interferometric SAR image quality, the interferometric phase, and the coherent coefficient. The research provides theoretical basis for the MOCO precision in signal processing of the airborne high precision and airborne repeat-pass interferometric SARs.

Synthetic aperture radar autofocus based on time-frequency transform

Phase gradient autofocus (PGA) is a robust tool for high resolution synthetic aperture radar (SAR) phase estimation and correction under the assumption that some strong scatterers with high signal-to-clutter ratio (SCR) are available within the unfocused image. In this paper, an advanced PGA algorithm is proposed, where the short-time Fourier transform (STFT) firstly is used to estimate the quadratic phase error (QPE) and the PGA is applied with a more effective filtering in the short-time Fourier domain (STFD). The proposed algorithm can relax the limitation on the scene content and achieve well-focused images without iteration. The validity of the proposed algorithm is demonstrated with the experimental results of simulation data and real radar data.

Auto-registration imaging algorithm for airborne dual-antenna interferometric SAR

In order to reduce the complex image registration process for the airborne dual-antenna interferometric SAR (InSAR) system, a new auto-registration imaging algorithm is presented in the paper, which implements the registration step directly at the SAR raw data processing stage based on the EOK algorithm and the ECS algorithm. For the mismatch caused by the path difference of two antennas, it realizes the precision registration of interferometric image in the range by means of high degree polynomial approximation using scaling principle and shifting compensation in course of the imaging procedure. Finally, through the processing of the real interferometric data, better interferograms are obtained and the validity of the algorithm proposed in the paper is proved.