IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 14, NO. 11, NOVEMBER 2017 2013
Recent Advances in Synthetic Aperture Radar Remote Sensing—Systems, Data
Processing, and Applications
Hongbo Sun,Senior Member, IEEE, Masanobu Shimada,Fellow, IEEE, and Feng Xu, Senior Member, IEEE
Abstract— This letter closes a special stream consisting of selected papers from the fifth Asia–Pacific Conference on Syn- thetic Aperture Radar in 2015 (APSAR 2015). The latest research results and outcomes from APSAR 2015, particularly on the synthetic aperture radar (SAR) systems/subsystems design, data processing techniques, and various SAR applications in remote sensing, are summarized and presented. All these results repre- sent the recent advances in SAR remote sensing. Hopefully, this letter can provide some references for SAR researchers/engineers and stimulate the future development of SAR technology for remote sensing.
Index Terms— Radar imaging, remote sensing, signal process- ing, synthetic aperture radar (SAR).
I. INTRODUCTION
S
INCE the launch of the first civilian synthetic aperture radar (SAR) satellite—SEASAT—in 1978, SAR systems have been widely used in the earth remote sensing for nearly 40 years. Today, more than 20 spaceborne SARs and even more airborne SARs are being operated by many coun- tries for research and operational missions. All these sys- tems provide numerous high-resolution SAR data for various remote sensing applications, including geoscience and climate change research, environmental and earth system monitoring, 2-D/3-D surface mapping, and security-related applications such as ground moving target detection and tracking. These systems and data resources greatly facilitate the research and development of SAR technology and its applications in remote sensing.The Asia–Pacific Conference on Synthetic Aperture Radar (APSAR) is a series of international conferences devoted to SAR technology development and applications.
With the dramatic increase in the number of operational spaceborne/airborne SAR systems in recent years, there is now high demand for more gatherings, where the regional and international SAR and radar communities can share their tech- nical expertise. The APSAR provides an opportunity for SAR engineers and researchers to present the latest research and development outcomes and discuss with colleagues from all
Manuscript received August 5, 2017; accepted August 28, 2017. Date of publication September 26, 2017; date of current version October 25, 2017.
(Corresponding author: Hongbo Sun.)
H. Sun is with the Temasek Laboratories, Nanyang Technological Univer- sity, Singapore 637553 (e-mail: [email protected]).
M. Shimada is with the School of Science and Engineering, Tokyo Denki University, Tokyo 350-0394, Japan (e-mail: [email protected]).
F. Xu is with the Key Laboratory for Information Science of Electro- magnetic Waves (MoE), Fudan University, Shanghai 200433, China (e-mail:
Digital Object Identifier 10.1109/LGRS.2017.2747602
over the world, especially the Asia–Pacific region. Since 2007, five APSAR conferences have been successfully held in every odd year in Asia–Pacific countries such as China, South Korea, Japan, and Singapore (complementary to the EUSAR con- ferences in every even year in European countries). APSAR 2015 is the fifth APSAR which was held in Marina Bay Sands, Singapore, during September 1–4, 2015. In total, 317 delegates from 23 countries attended this conference and 205 high- quality technical papers were presented (141 papers in oral presentations and 64 papers in poster presentations), which represent the recent developments of SAR technologies and applications for remote sensing.
In view of the success of APSAR 2015, IEEE GEOSCIENCE AND REMOTESENSINGLETTERSopened a call for submis- sion of research papers, through a special stream devoted to reporting the latest research results and outcomes from APSAR 2015 papers. In total, 61 submissions were received, and after rigorous peer review, 26 papers were accepted for publication. This letter concludes this special stream by sum- marizing the main contributions of those papers. The rest of this letter is organized as follows. Section II introduces a few novel concepts for SAR system/subsystem design. Section III presents some advanced SAR data processing techniques.
Section IV presents some new results of SAR applications in remote sensing. Finally, Section V concludes this letter.
II. SAR SYSTEM/SUBSYSTEM
In the past decades, the SAR system has evolved very fast, synchronized with the progress of the hardware and digital technologies and signal processing theory. By the 1990s, the first generation of the L-/C-/X-band SAR systems were experimentally operated by several countries and leaned on the balance between the technology status and the user require- ments. As for applying the SAR to environmental monitoring, the requirements on the SAR system are focused on the high resolution, wide swath, high sensitivity (including polariza- tion), lightweight, and compactness. However, there are several limitations affecting the SAR system performance. In recent years, the SAR has been investigated and designed with the aim of breaking the limitation down and expanding its func- tionalities under the given conditions: wider swath, lower data rate, errorless processing, information enrichment, etc. Five papers have been published under this special stream, which presented the latest development in the fields of frequency- modulated continuous-wave (FMCW) SAR, geosynchronous SAR (Geo-SAR), data compression, wideband dual-polarized
1545-598X © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2014 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 14, NO. 11, NOVEMBER 2017
SAR antenna, and multiswath SAR acquisition. The contribu- tions of these papers are summarized as follows.
Villano et al.[1] proposed a data volume reduction tech- nique which is applicable to the staggered SAR acquisition mode—a future high-resolution wide-swath SAR. This tech- nique allows a significant reduction of the data volume for the systems employing a pulse repetition frequency (PRF), which is much larger than the processed bandwidth, with negligible degradation of the azimuth ambiguity. It is currently consid- ered, together with the staggered SAR mode, for Tandem-L system.
Huang et al.[2] developed a wideband dual-polarized and dual-monopulse compact array for SAR system integration applications. An advanced 3-D metal-direct-printing technique was utilized for the array fabrication to improve the hard- ware performance, especially in terms of efficiency. Experi- mental results showed an excellent monopulse performance over 12.5% operational bandwidth. In particular, up to 95%
efficiency was obtained at the center frequency in the sum patterns. In addition, the measured null-depth and amplitude imbalance levels are−30 and 0.3 dB, respectively, in the dif- ference patterns. This sophisticated design can be very useful to customize for different practical SAR system requirements and applications.
Geo-SAR could be the next main SAR system in space for earth observation, since it is capable of providing wide swath coverage and high repeat-rate observations. Chen et al. [3]
investigated the effects of signal bandwidth and integration time parameters for each orbit position. They were optimized to have a circular ground impulse response width (IRW) as close as possible. To solve the problem of varying optimized signal bandwidth with satellite position, a parameter opti- mization model was built for obtaining an optimal signal bandwidth for the entire orbit. Subsequently, the corresponding length of integration time for the entire orbit was calculated.
Simulation results for both circular and elliptic orbits validated the proposed extraction method of the ground IRW curve and the parameter optimization model.
The FMCW SAR possesses the features of higher res- olution, lightweight, low cost, and low power dissipation.
Gu et al. [4] proposed a new distortion correction method for FMCW SAR real-time imaging, which conquered the real-time processing of the range deviations from the ideal path in a way that the rotation of the segmented images was corrected using a simple multiplication of the phase rotation and the sample shift in range and azimuth. Compared to the conventional fast Fourier transform-based processing technique, the proposed method can reduce the processing time to 1/500. The experiments validated that time consumption can be dramatically reduced from 0.3 s to only 0.717 ms for an image with 800-m swath.
An SAR time sharing, which assigns incoming pulses to different areas for the N-times-increased PRF, allows the acquisition of different strip images either to increase azimuth resolution or to have a multi-image system without increasing the number of receivers or partitioning the antenna.
This method is called the DIscrete Stepped Strip (DI2S) tech- nique, and is introduced and analyzed by Calabreseet al.[5].
This innovative method is actually patent pending and can be conceived to satisfy many applications in both crisis situations and ordinary operations, allowing the improvement and the flexibility of system capability.
III. SAR DATAPROCESSINGTECHNIQUES
Together with the development of SAR systems/subsystems, the SAR data processing techniques are also developing rapidly to adapt for some new or special working modes and improve the image quality or target detection performance.
First, in addition to the normal stripmap, spotlight, and scanSAR modes, some special SAR working modes have attracted more and more attention of international SAR com- munity in recent years. For example, the squinted SAR is capable of observing the specific region of interest several times to generate multiple acquisitions during a single pass by properly adjusting the squint angle of the antenna, which is often desired in spaceborne SAR applications. Chen et al.[6]
proposed a novel imaging algorithm for focusing the space- borne squinted sliding-spotlight SAR data, in which a modified accurate chirp scaling kernel was derived to realize range com- pensation, and an azimuth spatial variation removing method based on the principle of nonlinear chirp scaling was proposed to equalize the Doppler rates of the targets located at the same range cell. With this algorithm, the high-resolution images can be formed for spaceborne squinted sliding-spotlight mode.
Other than the dedicated SAR satellites, there is also increasing interest in the bistatic SAR concept using global navigation satellite systems (GNSSs) as illuminators, since the GNSSs have a large number of constellations and can provide continuous illumination for very large areas with a short revisit time. Unfortunately, the image quality obtained by such systems still cannot be compared with that of the dedicated SAR satellite systems due to the low power density of the GNSS signal. Zenget al.[7] proposed a coherent fusion method to jointly process the repeat-pass images acquired by using GNSSs to enhance the final image quality. In the experi- mental validation, 22 days repeat-pass data were collected with Beidou-2/Compass-2 satellites and the final image quality was improved considerably.
Nowadays, the unmanned aerial vehicle (UAV) is widely used in both military and civilian applications. It also offers a low-cost platform for SAR remote sensing. Unfortunately, due to its small size and lightweight, the UAV (particularly, the consumer-grade UAV) often suffers from significant tur- bulence and cannot maintain a straight path during the flight, which bring many technical challenges to the SAR image formation. To solve this problem, Yanget al.[8] proposed a new spectrum-oriented fast factorized back-projection (FFBP) algorithm for SAR imaging on maneuvering platforms with arbitrary trajectory. Different from the conventional FFBP algorithms developed in polar grid, the proposed algorithm was derived in quasi-polar gird and the motion-induced phase error was handled as a space-invariant component, which greatly facilitates the phase autofocusing process during FFBP recursions.
Ground moving target indication (GMTI) is one of the most important tasks for SAR systems. To suppress the ground
SUNet al.: RECENT ADVANCES IN SAR REMOTE SENSING—SYSTEMS, DATA PROCESSING, AND APPLICATIONS 2015
clutter and detect the weak moving targets, conventional techniques such as displaced phase center antenna, space–time adaptive processing, and along-track interferometry (ATI) have been proposed and used in many SAR systems. Under this spe- cial stream, Liet al.[9] proposed an efficient GMTI approach by combining ATI detection and go decomposition algorithm, where an ATI robust principal component analysis method is used in the predetection to decrease the probability of missing alarm, and a novel magnitude and phase detector is used in the postdetection to reduce the probability of false alarm, and overall a very robust moving target detection performance can be achieved. Xuet al.[10] proposed another GMTI technique based on the shadow feature of targets, in which the geometric relationships between the moving object and its shadow in position and size were analyzed, and an efficient shadow detection method based on multifeature fusion was presented.
Unlike the conventional GMTI techniques, this shadow-aided method does not depend on good clutter suppression and can be a promising GMTI technique in the future.
The demand for higher radar image resolution has been higher in recent years, which has led to a rapid increase in the data volume recorded by SAR system under the Nyquist sam- pling theorem. To reduce the total data volume as well as the radar resources (e.g., spectrum occupation) while maintaining good radar image quality, the compressive sensing techniques have been widely considered for SAR or inverse SAR (ISAR) imaging by many researchers. Chen et al. [11] proposed a genetic algorithm-based measurement matrix optimization technique for the ISAR imaging with the stepped-frequency radar, in which the target characteristics and actual physical observation process were utilized for the measurement matrix optimization with the objectives of saving radar frequency resource and reducing the overall data volume. On the other hand, considering the sparse characteristic of the target image, the compressive sensing techniques can also be used to improve the SAR/ISAR image quality. Sunet al.[12] proposed a joint sparsity-based technique to remove the frequency modulation effects of micromotion (i.e., the mechanical rota- tion or vibration of some structures on the target main body) and enhance the image quality of target main body itself, which was validated by experimental ISAR data.
IV. SAR APPLICATIONS INREMOTESENSING
With the rapid advancement of SAR technologies, SAR image interpretation and information retrieval are the crit- ical final steps for successful application of the numerous SAR sensors and systems. Different from the optical image, the SAR image is usually difficult for human to understand.
Nevertheless, the SAR image contains rich information about the imaged scene or target, e.g., geometry, material, and structure. This section briefly reviews the application-oriented research progresses reported in this special stream.
New SAR technologies lead to high-resolution, multidimen- sion, and multimode SAR data, which have brought many challenges in image interpretation. It has to be rooted at the understanding of fundamental electromagnetic (EM) scattering mechanisms. SAR advanced information retrieval is proposed, which couples advanced information processing methods at
the mathematical side with EM scattering theories at the physics side. Xu et al. [13] proposed a framework for scene reconstruction, which aims to transfer the SAR image to human-understandable representation of man-made targets and natural environment. It includes three key elements: 1) a dictio- nary of parametric scatterer model; 2) a method for scatterer recognition and parameter estimation; and 3) a method for target reconstruction. A preliminary case is presented, where the simulated 3-D SAR image of a simple target is successfully reconstructed to a solid geometry.
Interferometry is, without a doubt, one of the most success- ful applications of SAR technology. The spaceborne interfer- ometric SAR (InSAR) has become an indispensable tool for subsidence monitoring. Sumantyoet al.[14] reported a study of using ALOS PALSAR to investigate the land deformation at 11 watersheds of the West Java Mega Urban Region.
It revealed that land deformation at the Bandung city area has a significant impact on sedimentation velocity along the eastern Jakarta strait. Another study by Wu and Hu [15] employed the TerraSAR-X data to monitor the ground subsidence along the Shanghai maglev railway. Permanent scatterers near the maglev line were studied and an average subsidence rate
of<3 mm/a was observed and verified. In addition, they
found that some ground subsiding spots have subsidence rates of more than 10 mm/a. Another letter by Cerchielloet al. [16]
reported a study of predicting building damage due to subsi- dence using InSAR time series in conjunction with a semiem- pirical model of buildings. The given case study generated a building damage risk map of the southern part of the city of Rome. Wuet al.[17] approached building damage evaluation after earthquake using the new TerraSAR-X starring spotlight mode. The Beichuan County damaged by the May 12, 2008 Wenchuan earthquake was selected as the research area.
Features such as backscattering and texture measurements were analyzed. It concluded that the new starring spotlight mode has the potential for building damage detection and discrimination of basic damage classes.
SAR polarimetry is another major application.
Kimet al.[18] applied the polarimetric SAR (PolSAR) data to the paddy field mapping. They presented a simple rice field mapping method in the late vegetative stage using both the topographic feature and the scattering feature of the PolSAR data. Topographic features including the polarization orienta- tion angle and the dominant beta angle from the PolSAR data were analyzed with respect to the slope from a digital elevation model. Scattering features including the copolarized ratio and the copolarized phase difference were combined to map only paddy fields, of which scattering is dominated by the ground- canopy double-bounce scattering. The C-band RADARSAT-2 data were also used in this letter to demonstrate the efficacy of their method. Guinvarc’h and Thirion [19] analyzed the cross-polarization amplitudes of obliquely oriented buildings and found their potential applications in urban remote sensing. Sukawattanavijitet al. [20] presented a study of using a genetic algorithm and support vector machine to improve the land mapping using the fusion of SAR and optical images. The data used include multifrequency RADARSAT-2 SAR images and THAICHOTE multispectral images.
2016 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 14, NO. 11, NOVEMBER 2017
SAR automatic target recognition (ATR) is another appli- cation interesting to many users. In this special stream, two papers reported SAR ATR studies. Dai et al. [21] proposed a modified constant false alarm rate (CFAR) algorithm based on object proposals for ship target detection. This letter addressed the multiscale target detection issue with an adaptive CFAR window size generated by object proposal. Sun et al. [22]
proposed an SAR ATR algorithm based on dictionary learning and the joint dynamic sparse representation. It first learns dictionaries from the features extracted from the SAR image and then classifies targets with sparse representation.
V. CONCLUSION
As predicted by many researchers and engineers in the mid-2000s, SAR has been taking a walk in the golden age where the number of the SAR sensors is dramatically increased and the functionalities are widely expanded. This benefits from the reasonable matching of the sensor performance and the needs from the society—the human society to the engineering and scientific societies. In that, the SAR is directing to the higher performance of wider swath, higher repeatability, lighter weight, higher sensitivity, lower data rate under the assured image quality, and the linkage to the measurable geophysical parameters. One of the typical reasons for higher needs is to quickly respond to the change of the earth and environment. For this objective, the APSAR 2015 was held in Singapore in 2015 and achieved great success. This special stream titled “Recent Advances in the Synthetic Aperture Radar Remote Sensing—Systems, Data Processing, and Appli- cations” has taken up the essence of the APSAR 2015 in Singapore. It can be promising for the SAR researchers and engineers to taste the richest part of the recent progress in technology in this special stream.
REFERENCES
[1] M. Villano, G. Krieger, and A. Moreira, “Onboard processing for data volume reduction in high-resolution wide-swath SAR,” IEEE Geosci.
Remote Sens. Lett., vol. 13, no. 8, pp. 1173–1177, Aug. 2016.
[2] G.-L. Huang, S.-G. Zhou, T.-H. Chio, C.-Y.-D. Sim, and T.-S. Yeo,
“Wideband dual-polarized and dual-monopulse compact array for SAR system integration applications,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 8, pp. 1203–1207, Aug. 2016.
[3] J. Chen et al., “A parameter optimization model for geosynchronous SAR sensor in aspects of signal bandwidth and integration time,”IEEE Geosci. Remote Sens. Lett., vol. 13, no. 9, pp. 1374–1378, Sep. 2016.
[4] C. Gu, W. Chang, X. Li, G. Jia, and X. Luan, “A new distortion correction method for FMCW SAR real-time imaging,”IEEE Geosci.
Remote Sens. Lett., vol. 14, no. 3, pp. 429–433, Mar. 2017.
[5] D. Calabrese, V. Mastroddi, and S. Federici, “DI2S multiswath innova- tive technique for SAR acquisitions optimization,”IEEE Geosci. Remote Sens. Lett., vol. 14, no. 10, pp. 1820–1824, Oct. 2017.
[6] J. Chen, H. Kuang, W. Yang, W. Liu, and P. Wang, “A novel imaging algorithm for focusing high-resolution spaceborne SAR data in squinted sliding-spotlight mode,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 10, pp. 1577–1581, Oct. 2016.
[7] T. Zeng, T. Zhang, W. Tian, and C. Hu, “Space-surface bistatic SAR image enhancement based on repeat-pass coherent fusion with BeiDou-2/Compass-2 as illuminators,”IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1832–1836, Dec. 2016.
[8] L. Yang, L. Zhao, S. Zhou, G. Bi, and H. Yang, “Spectrum-oriented FFBP algorithm in quasi-polar grid for SAR imaging on maneuvering platform,”IEEE Geosci. Remote Sens. Lett., vol. 14, no. 5, pp. 724–728, May 2017.
[9] J. Li, Y. Huang, G. Liao, and J. Xu, “Moving target detection via efficient ATI-GoDec approach for multichannel SAR system,”
IEEE Geosci. Remote Sens. Lett., vol. 13, no. 9, pp. 1320–1324, Sep. 2016.
[10] H. Xu, Z. Yang, G. Chen, G. Liao, and M. Tian, “A ground moving target detection approach based on shadow feature with multichannel high-resolution synthetic aperture radar,” IEEE Geosci. Remote Sens.
Lett., vol. 13, no. 10, pp. 1572–1576, Oct. 2016.
[11] Y.-J. Chen, Q. Zhang, Y. Luo, and Y.-A. Chen, “Measurement matrix optimization for ISAR sparse imaging based on genetic algorithm,”
IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1875–1879, Dec. 2016.
[12] L. Sun, X. Lu, and W. Chen, “Joint sparsity-based ISAR imaging for micromotion targets,”IEEE Geosci. Remote Sens. Lett., vol. 13, no. 11, pp. 1734–1738, Nov. 2016.
[13] F. Xu, Y.-Q. Jin, and A. Moreira, “A preliminary study on SAR advanced information retrieval and scene reconstruction,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 10, pp. 1443–1447, Oct. 2016.
[14] J. T. S. Sumantyo et al., “Analysis of coastal sedimentation impact to Jakarta Giant Sea Wall using PSI ALOS PALSAR,”IEEE Geosci.
Remote Sens. Lett., vol. 13, no. 10, pp. 1472–1476, Oct. 2016.
[15] J. Wu and F. Hu, “Monitoring ground subsidence along the Shanghai Maglev zone using TerraSAR-X images,” IEEE Geosci. Remote Sens.
Lett., vol. 14, no. 1, pp. 117–121, Jan. 2017.
[16] V. Cerchiello, G. Tessari, E. Velterop, P. Riccardi, M. Defilippi, and P. Pasquali, “Building damage risk by modeling interferometric time series,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 4, pp. 509–513, Apr. 2017.
[17] F. Wu, L. Gong, C. Wang, H. Zhang, B. Zhang, and L. Xie, “Sig- nature analysis of building damage with TerraSAR-X new staring SpotLight mode data,”IEEE Geosci. Remote Sens. Lett., vol. 13, no. 11, pp. 1696–1700, Nov. 2016.
[18] Y. Kim, J. Oh, D.-J. Kim, and Y. Kim, “Paddy field mapping using topographic and scattering features of PolSAR data,” IEEE Geosci.
Remote Sens. Lett., vol. 14, no. 4, pp. 484–488, Apr. 2017.
[19] R. Guinvarc’h and L. Thirion, “Cross-polarisation amplitudes of obliquely oriented buildings with application to urban areas,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 11, Nov. 2017.
[20] C. Sukawattanavijit, J. Chen, and H. Zhang, “GA-SVM algorithm for improving land-cover classification using SAR and optical remote sensing data,” IEEE Geosci. Remote Sens. Lett., vol. 14, no. 3, pp. 284–288, Mar. 2017.
[21] H. Dai, L. Du, Y. Wang, and Z. Wang, “A modified CFAR algorithm based on object proposals for ship target detection in SAR images,”
IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1925–1929, Dec. 2016.
[22] Y. Sun, L. Du, Y. Wang, Y. Wang, and J. Hu, “SAR automatic target recognition based on dictionary learning and joint dynamic sparse representation,” IEEE Geosci. Remote Sens. Lett., vol. 13, no. 12, pp. 1777–1781, Dec. 2016.