This PDF file contains the front matter associated with SPIE Proceedings Volume 12732, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Remote sensing scene classification has been extensively studied for its critical roles in geological survey, oil exploration, traffic management, earthquake prediction, wildfire monitoring, and intelligence monitoring. In the past, the Machine Learning (ML) methods for performing the task mainly used the backbones pretrained in the manner of supervised learning (SL). As Masked Image Modeling (MIM), a self-supervised learning (SSL) technique, has been shown as a better way for learning visual feature representation, it presents a new opportunity for improving ML performance on the scene classification task. This research aims to explore the potential of MIM pretrained backbones on four well-known classification datasets: Merced, AID, NWPU-RESISC45, and Optimal-31. Compared to the published benchmarks, we show that the MIM pretrained Vision Transformer (ViTs) backbones outperform other alternatives (up to 18% on top 1 accuracy) and that the MIM technique can learn better feature representation than the supervised learning counterparts (up to 5% on top 1 accuracy). Moreover, we show that the general-purpose MIM-pretrained ViTs can achieve competitive performance as the specially designed yet complicated Transformer for Remote Sensing (TRS) framework. Our experiment results also provide a performance baseline for future studies.

The lack of stable and coherent natural targets can threat the effectiveness of Interferometric Synthetic Aperture Radar (InSAR) applications. To overcome this issue, active and passive radar reflectors are designed. Thanks to their low cost of construction and maintenance, passive radar reflectors are even more employed as coherent targets to assess potential displacement measurements of land, buildings, and infrastructures. In the present study, different types of passive radar reflectors are investigated by simulating their backscattering characteristics through a 3D electromagnetic software and by calculating their radar cross sections at different azimuth and incidence angles. Simulation results have been examined by considering the characteristics of current SAR satellite missions orbiting on different planes (Sun-Synchronous Orbit/Mid-Inclination Orbit) and considering both passes (Ascending/Descending), as well as by analyzing different orientations of SAR antenna (i.e., Right/Left look sides and incidence angles). Advantages and disadvantages of the investigated passive radar reflectors are highlighted in terms of their visibility on multiple Line Of Sights (LOS). Two carrier frequencies have been selected, that are close to those of operational SAR satellites: 5.405 GHz (C-band) and 9.66 GHz (X-band).

This work is devoted to experiment the use of artificial reflectors, both passive and active, for supporting the displacement measurements derived through multi-temporal SAR interferometry (MTInSAR). An experimental site has been set up by deploying two corner reflectors (CRs) and three active reflectors. Each CR consists of three triangular metal panels whose internal leg is respectively 69.5 cm (CR0) and 1.05 m (CR1), welded perpendicularly to each other to form a trihedral shape. Concerning the active reflectors, one is a C-band Electronic Corner Reflector (ECR-C), which is compatible with operation frequency of Sentinel-1, while the other two are Active Radar calibrators (ARC) designed in the early ‘90s for supporting the calibration of SAR sensors operating at C-band (ARC-C) and X-band (ARC-X). ECR-C and ARC-C have been tuned to work with the Sentinel-1 C-band SAR mission, while ARC-X has been exploited for working with COSMO-SkyMED X-band constellation. Ascending and descending data acquired by COSMO-SkyMED and Sentinel-1 have been processed through the SPINUA MTInSAR algorithm, then time series of both SAR amplitude and displacement values have been analysed for comparing results from CRs of different sizes and active reflectors. Generally, we found a good agreement between the measured and the nominal expected backscattered signals (both in amplitude and phase). The CR size impacts on the signal quality: CR0 is smaller and so noisier and more exposed to interaction with both ERC-C and urban structures than CR1. Phase stability of active reflectors is sometimes below the expected value. This instrumented test site will be proposed as an open experimental site by providing open data for testing and validating interferometric techniques.

FMCW airborne SAR can collect data from different viewing angles, and can offer rapidly informative imagery of strategic infrastructure, or disaster areas at local scales. As they transmit low power, they can be mounted on light platforms. In 2022 we conducted a measurement campaign with the FHR’s MIRANDA35 system. It allows 1) topographic mapping by means of a tomographic configuration, 2) along-track interferometry for air and ground moving target indication, and 3) polarimetry for target classification and recognition. We show the performance of the system, and diverse level-1 and level-2 products, such as multi-aspect digital elevation models.

The detection of Radio frequency interference (RFI) is an important work for synthetic aperture interferometric radiometer (SAIR). However, the previously proposed RFI detection methods have difficulties in dealing with complex RFI scenes with large dynamic RFI intensity. Because the strong RFI signal energy picked up by the sidelobes of synthetic beam (or array factor) of SAIR are easily detected as false positive RFIs, resulting in false alarm phenomenon. To address this problem, we propose an improved RFI source detection method based on the array factor property (AFP) of SAIR. The AFP-based RFI detection (AFPRD) method mainly consists of three steps. First, RFI sources are recovered from visibility function samples through sparse reconstruction method reweighted l1-norm minimization (RL1). Second, the AFP of SAIR is analyzed concerning distribution characteristics of main beam and sidelobes. Since the recovered RFI map can be regarded as the convolution of the actual RFI map and AF, the relative positions of the strong RFI and its resulting false positive RFIs are consistent with the relative position of the main lobe and the sidelobes in AF. Based on the AFP analysis, we present a new spatial weight indicator (SWI) to describe the probability of one possible RFI being false positive. And each pixel in the RFI map is assigned with a SWI value to generate the SWI-based probability matrix (SPM). Third, the SPM and the SR-based map are combined to reconstruct the RFI image, where false positives are filtered out while true RFI sources are retained. In the experiment, the validity of the proposed method is verified by synthetic data and real data measured by SMOS satellite.

The Soil Moisture and Ocean Salinity (SMOS) mission led by the European Space Agency (ESA) is aimed at globally monitoring the Earth surface moisture and ocean salinity. As the single payload of the SMOS satellite, the Microwave Interferometric Radiometer with Aperture Synthesis (MIRAS) operates in the protected L-band. Nevertheless, the artificial sources emitting close to or/and entirely in this band are contaminating the collected remote sensing data and deteriorating the performance of the SMOS mission. In this article, we propose a method based on interpolation-assisted matrix completion (IMC) for the localization of RFI sources. The method firstly constructs the augmented covariance matrix (ACM). Secondly, it exploits the local property of visibility function sampling data in the spatial-frequency domain and adopts a local polynomial regression model to interpolate missing visibility data. Thirdly, it exploits the low-rank property of the ACM and recovers this ACM via the matrix completion technique. Finally, the subspace-based algorithm (i.e., MUSIC) is used to estimate source direction-of-arrivals (DOAs) for RFI localization. The proposed RFI localization method is termed as IMC-MUSIC. It is mainly devised for application scenarios where there exist data loss due to inherent array geometry constraint or potential hardware failure and compressed measurements are adopted for reducing system complexity. Validation results show that compared with traditional methods such as discrete Fourier transform (DFT) inversion and covariance-based DOA estimation, the proposed method shows better localization accuracy for different data loss or compression ratios.

Change detection in highly dynamic areas, such as vegetated areas, is necessary to improve the thematic accuracy of national land use/cover map production. Sentinel-1 data have proved its potential for change detection. However, its suitability in mountainous and sparse forested areas is not well known. Here, we propose a straightforward clear-cut detection method, using monthly backscatter and coherence composites in both polarizations. The monthly clear-cut masks are estimated using a multivariate alteration detection (MAD) algorithm, using the previous and following composite, and a threshold value corresponding to the 98th percentile. The method was applied to a test site in northern Portugal, mainly characterized by the presence of dense eucalyptus forest, in a mountainous area. A Sentinel-1 time series from February to October 2018 was considered. Results showed that the monthly clear-cut estimation is not only highly influenced by rainfall events, with F1-score values less than 0.11, for the rainy season, but also by terrain-induced geometric distortions and foliage characteristics of the eucalyptus stands. The highest accuracy metrics were obtained for June, with an F1-score value of 0.45, which was still considered unfavorable. Inaccurate monthly estimations are also affected by clear-cut events that occur over more than one month, with the sum of all masks producing results with F1- scores less than 0.41. When applied to another region, located southwest of the former site and near the coastline, but characterized by smoother slopes and scattered and sparse forested stands, the method retrieved equivalent recall, but increased precision of around 0.78 (false alarms reduction), resulting in higher F1-score values (0.56).

In this study, we apply time-series interferometric synthetic aperture radar (InSAR) technique to Korea multi-purpose satellite-5 (KOMPSAT-5) images to analyze surface deformation. Experimental results obtained using the small baseline subset (SBAS) technique expressively represent large crucial movements due to volcanic eruption in 2018 as well as long-period surface deformation trends.

The Republic of Korea is presently operating the KOMPSAT-5 Synthetic Aperture Radar (SAR) satellite and intends to launch the KOMPSAT-6 SAR satellite and a constellation system of micro-SAR satellites in the future. Therefore, the demand for SAR satellite information is increasing in various fields, but many public and private users who use national satellite information in practice use Electro-Optical and Infrared (EO/IR) satellite images. In this research, an independent KOMPSAT-5 SAR imagery analysis tool for Automatic Change Detection and Cueing (ACDC) was developed to promote the utilization of domestic SAR satellite information by practitioners in public sector organizations in Korea. For the development of SAR ACDC, the functions required for each processing step were analyzed in detail, and this was implemented as a proto-type tool. Algorithms that are the basis of the KOMPSAT-5 ACDC, for example, KOMPSAT-5 SAR image import, metadata comparison, SAR noise removal filtering, image co-registration, and change image generation, etc. were implemented. Moreover, the proto-type tool was developed so that users can use it correctly and conveniently by supporting an actual user-friendly interface. The ACDC algorithms and the tool are anticipated to be used in the future by users who use Korea's national satellite data for early detection of illegal ships and structures, as well as confirmation of wide-area disaster damage.