Dissertations / Theses on the topic 'Ground based synthetic aperture radar'

This thesis describes a first hands-on experience working with a Ground-Based Synthetic Aperture Radar (GB-SAR) at the Institute of Geomatics in Castelldefels (Barcelona, Spain), used to exploit radar interferometry usually employed on space borne platforms. We describe the key concepts of a GB-SAR as well as the data processing procedure to obtain deformation measurements. A large part of the thesis work have been devoted to development of GB-SAR processing tools such as coherence and interferogram generation, automating the co-registration process, geocoding of GB-SAR data and the adaption of existing satellite SAR tools to GB-SAR data. Finally a series of field campaigns have been conducted to test the instrument in different environments to collect data necessary to develop GB-SAR processing tools as well as to discover capabilities and limitations of the instrument.   The key outcome of the field campaigns is that high coherence necessary to conduct interferometric measurements can be obtained with a long temporal baseline. Several factors that affect the result are discussed, such as the reflectivity of the observed scene, the image co-registration and the illuminating geometry.
Det här examensarbetet bygger på erfarenheter av arbete med en mark-baserad syntetisk apertur radar (GB-SAR) vid Geomatiska Institutet i Castelldefels (Barcelona, Spanien). SAR tekniken tillåter radar interferometri som är en vanligt förekommande teknik både på satellit och flygburna platformar. Det här arbetet beskriver instrumentets tekniska egenskaper samt behandlingen av data for att uppmäta deformationer. En stor del av arbetet har ägnats åt utveckling av GB-SAR data applikationer som koherens och interferogram beräkning, automatisering av bild matchning med skript, geokodning av GB-SAR data samt anpassning av befintliga SAR program till GB-SAR data. Slutligen har mätningar gjorts i fält for att samla in data nödvändiga for GB-SAR applikations utvecklingen samt få erfarenhet av instrumentets egenskaper och begränsningar.   Huvudresultatet av fältmätningarna är att hög koherens nödvändig för interferometriska mätningar går att uppnå med relativ lång tid mellan mätepokerna. Flera faktorer som påverkar resultatet diskuteras, som det observerade områdets reflektivitet, radar bild matchningen och den illuminerande geometrin.

This thesis works through the design of a radar-based system for imaging snowpacks remotely and over large areas to assist in avalanche prediction. The key to such a system is the ability to image volumes of snow at shallow, spatially-varying angles of incidence. To achieve this prerequisite, the design calls for a ground-based Synthetic Aperture Radar (SAR) capable of generating three-dimensional, ultra-high-resolution images of a snowpack. To arrive at design parameters for this SAR, the thesis works through relevant principles in avalanche mechanics, alpine-snowpack geophysics, and electromagnetic scattering theory. The thesis also works through principles of radar, SAR, antenna, and image processing theory to this end. A preliminary system is implemented to test the feasibility of the overall design. The preliminary system demonstrates ultra-high-resolution, three-dimensional imaging capabilities and the ability to image the volume of multiple alpine snowpacks. Images of these snowpacks display the structural patterns indicative of different layers in the snowpacks. Possible attributions of the patterns to physical properties in the snowpack are explored, but conclusions are not arrived at. Finally, lessons from the implementation of this preliminary system are discussed in terms of opportunities to be capitalized upon and problems to be overcome in future systems that more faithfully realize the complete design set forth in the thesis.

Le tecniche Multi-temporali SAR interferometriche (Mt-InSAR) rappresentano oggigiorno strumenti consolidati per mappare l’evoluzione temporale dei fenomeni di deformazione del suolo Terrestre. Queste tecniche utilizzano congiuntamente sets di interferogrammi SAR differenziali al fine di estrarre la componente legata alla deformazione e produrre così serie storiche di deformazione dei bersagli osservati dal sensore. L'affidabilità delle misure prodotte utilizzando algoritmi Mt-InSAR è strettamente legata alla capacità degli stessi algoritmi nell’isolare esclusivamente i segnali legati alla deformazione dal segnale complessivo interferometrico, e questa operazione diventa sempre più complessa all’aumentare dei livelli di rumore in ciascun interferogramma SAR coinvolto. Le tecniche Mt-InSAR canoniche sono altamente affidabili nel monitorare l'evoluzione dello spostamento dei target che risultano essere ampiamente stabili o coerenti per tutto il periodo di analisi. Diversamente, quando i bersagli sono particolarmente affetti da problemi di decorrelazione, le stime di deformazione ottenute risultano corrotte e inaffidabili. Questo pone le basi per lo sviluppo di processori Mt-InSAR avanzati che possano fornire stime accurate della deformazione del suolo anche in scenari con problemi di decorrelazione più o meno severi. In questo lavoro di tesi affronta dapprima lo studio dello stato dell’arte delle tecniche Mt-InSAR canoniche applicabili sia nel caso di piattaforme satellitare che terrestri, e dopodiché si propongono delle nuove tecniche Mt-InSAR per superare alcune delle criticità riscontrante. In particolare si studiano le tecniche convenzionali Mt-InSAR multigriglia per l'analisi dei target alla griglia di risoluzione spaziale più risoluta, evidenziandone le loro criticità in aree a media e bassa coerenza, e proprio in questo ambito è proposta una tecnica innovativa per meglio operare in ambienti decorrelati. Il metodo proposto si basa su efficienti operazioni con cui viene srotolata la fase (PhU) interferometrica eseguite alle scale spaziali native, ed in particolare, si srotolano dapprima gli interferogrammi alla scala di soluzione mediata (ML) attraverso algoritmi di PhU convenzionali (o avanzati). Successivamente, gli interferogrammi ML srotolati vengono utilizzati per facilitare le operazioni di PhU eseguite alla scala più fine (single-look). In dettaglio, gli interferogrammi multi-look srotolati vengono ricampionati alla griglia single-look e sottratti a modulo modulo-2π agli interferogrammi single-look. Gli interferogrammi epurati dai contributi a bassa frequenza vengono poi srotolati e aggiunti nuovamente agli interferogrammi multilook ricampionati alla griglia di risoluzione più fine. Per realizzare queste operazioni, a differenza dei metodi multigriglia canonici, non si utilizza alcun modello (lineare/non lineare) per recuperare le componenti di deformazione in alta frequenza. Infine, gli interferogrammi single-look srotolati sono opportunamente invertiti al fine di calcolare le serie storiche di deformazione del suolo attraverso un qualsiasi algoritmo a piccola baseline (SB) InSAR multi-temporale. I risultati sperimentali sono stati ottenuti elaborando una serie di dati SAR acquisiti dal sensore COSMO-SkyMed (banda X) sulla zona costiera di Shanghai, in Cina. La tesi prosegue analizzando le tecniche ai minimi quadrati pesate (WLS) e su come sono sfruttate nell’ambito InSAR al fine di migliorare l’operazione con cui si srotola la fase interferometrica e la generazione di serie storiche di deformazione. Proprio in questo contesto, utilizzando gli approcci WLS, si estende l'utilizzabilità dell'algoritmo Mt-InSAR Small BAseline Subset (SBAS) in aree caratterizzate da una coerenza spaziale medio-bassa. In particolare, pixel per pixel, si invertono esclusivamente le fasi interferometriche coerenti utilizzando una metrica a minimi quadrati pesati. Per cui attraverso una selezione adattiva, per ogni pixel si utilizzano ed invertono soltanto le fasi interferometriche coerenti, e tale caratterista può portare a diversi sottoinsiemi disgiunti di dati SAR, che sono poi invertiti sfruttando la Decomposizione a Valori Singolari Pesata (WSVD). Tuttavia, per taluni pixel, l’utilizzo esclusivo delle fasi interferometriche coerenti può portare in alcuni casi allo scarto di acquisizioni SAR particolarmente rumorose, il che si traduce in serie storiche di deformazione affidabili ma con campionamento temporale variabile. I risultati sperimentali sono stati condotti applicando la tecnica sviluppata ad un set di dati SAR acquisiti dai sensori COSMO-SkyMed (CSK) sulla regione Basilicata, nel sud Italia. Il lavoro di tesi continua analizzando le proprietà che ledono alla irrotazionalità delle triplette di fase di interferogrammi SAR multi-look. In particolare, si studiano le conseguenze delle incongruenze temporali di fase dei multi-look sulla generazione delle serie storiche di deformazione del suolo attraverso metodi SB Mt-InSAR. La ricerca condotta mostra come queste incongruenze di fase si possono propagare attraverso una rete temporale ridondante di interferogrammi SB, ed insieme agli errori di PhU, pregiudicano la qualità dei prodotti InSAR generati. In letteratura questo effetto va sotto il nome di bias di fase, il quale può pregiudicare l’affidabilità dei metodi SB quando si impostano delle soglie sulla massima baseline temporale troppo stringenti (nell’ordine di 30 giorni o meno). Proponiamo così, due nuovi metodi per la compensazione di tali fenomeni di bias, i quali metodi sono stati testati utilizzando dati SAR simulati e reali. I dati reali sono stati acquisiti dai sensori Sentinel-1A/B (banda C) sulle aree del Nevada (U.S.), e sulla zona del monte Etna in Sicilia, nel sud Italia. Dopo lo sviluppo di algoritmi per la parte satellitare, il lavoro si sposta sui sensori SAR terrestri (GB-SAR). In questo ambito proponiamo un metodo per stimare e compensare i disturbi introdotti dallo strato atmosferico (APS) in interferogrammi GB-SAR. Un’ambia analisi fisica, statistica e matematica dell'approccio presentato è fornita, discutendo inoltre le potenzialità e i limiti del metodo che a differenza di altri algoritmi, che stimano l'APS dai segnali di fase srotolati, nella metodologia proposta la compensazione avviene direttamente sul dato arrotolato, in modo tale che la stima non è affetta da nessun potenziale errore di PhU. Gli esperimenti eseguiti su dati InSAR GB-SAR simulati e reali confermano la validità della tecnica proposta, confermando inoltre che il metodo è vantaggioso nelle zone caratterizzate da una forte escursione di quota (come ad esempio nelle regioni Alpine e montuose). Infine, viene presentata un'applicazione SAR interferometrica per la stima delle deformazioni della superficie investigata in tre dimensioni (3-D) attraverso l'uso congiunto ed integrato di dati SAR acquisiti da piattaforme satellitari e terrestri. Più precisamente, la catena di combinazione interferometrica sviluppata si compone anche degli innovativi algoritmi Mt-InSAR sviluppati in questo lavoro di tesi, al fine di ottenere mappe di velocità media di deformazione 3-D direttamente alla griglia spaziale più risoluta possibile. Inoltre, in conclusione, vengono menzionate anche alcune interessanti applicazioni SAR satellitari in ambito di prevenzione ed analisi di particolari fenomeni naturali e indotti dall'uomo.
Multi-temporal SAR interferometric (Mt-InSAR) techniques are nowadays mature tools to measure the temporal evolution of the Earth’s surface with millimetric accuracy. The reliability of crustal measurements is closely related to the goodness of the used Mt-InSAR algorithms in isolating the deformation-related signal from the overall signal, and this becomes increasingly complex as the noise levels of each interferogram increase. Canonical techniques are highly reliable in monitoring the displacement evolution of targets that are found to be largely stable or coherent over the entire period of analysis. Otherwise, when the scatterers are particularly affected by decorrelation problems, the obtained deformation estimates turn out to be corrupted and unreliable. Thus, there is a strong demand for new advanced Mt-InSAR processors that can provide accurate estimates of crustal deformation even in scenarios with more or less severe decorrelation problems. This thesis work focuses on the study of multi-temporal InSAR techniques applicable in both satellite and terrestrial case. Specifically, the canonical Mt-InSAR multigrid techniques for analyzing targets at the finest resolution grid will be discussed extensively highlighting their criticality in medium to low coherence areas, and in this context an innovative technique is proposed to better operate in decorrelated environments. The new method relies on efficient phase-unwrapping (PhU) operations performed at the native spatial scales. In particular, a set of multi-look (ML) interferograms is first unwrapped using conventional (or advanced) PhU algorithms at the regional scale. Subsequently, ML unwrapped interferograms are used to facilitate the PhU operations performed at the local scale (single-look). Specifically, the unwrapped multi-look interferograms are resampled to the single-look grid and modulo-2π subtracted to the single-look interferograms. These phase residuals are then unwrapped and added back to the multi-look resampled interferograms. To accomplish these operations, at variance with alternative multiscale methods, no (linear/nonlinear) models are used to fit the spatial high-pass phase residuals. Finally, the unwrapped single-look interferograms are properly inverted to retrieve the ground displacement time series using any small baseline (SB)-oriented multitemporal InSAR tool. Experimental results are performed by processing a set of SAR data acquired by the X-band COSMO-SkyMed sensor over the coastal area of Shanghai, China. Then, the focusing moves on the Weighted Least-squares (WLS) techniques applied within the InSAR framework for improving the performance of the phase unwrapping operations as well as for better conveying the inversion of sequences of unwrapped interferograms to generate ground displacement maps. In both cases, the identification of low-coherent areas, where the standard deviation of the phase is high, is requested. Therefore, a WLS method that extends the usability of the Mt-InSAR Small BAseline Subset (SBAS) algorithm in regions with medium-to-low coherence is presented. In particular, the proposed method relies on the adaptive selection and exploitation, pixel-by-pixel, of the medium-to-high coherent interferograms, only, so as to discard the noisy phase measurements. The selected interferometric phase values are then inverted by solving a WLS optimization problem. Noteworthy, the adopted, pixel-dependent selection of the “good” interferograms to be inverted may lead the available SAR data to be grouped into several disjointed subsets, which are then connected, exploiting the Weighted Singular Value Decomposition (WSVD) method. However, in some critical noisy regions, it may also happen that discarding of the incoherent interferograms may lead to rejecting some SAR acquisitions from the generated ground displacement time-series, at the cost of the reduced temporal sampling of the data measurements. Thus, variable-length ground displacement time-series are generated. The presented experiments have been carried out by applying the developed technique to a SAR dataset acquired by the COSMO-SkyMed (CSK) sensors over the Basilicata region, Southern Italy. In the continuation of the thesis work, the properties characterizing the phase non-closure of multi-look SAR interferograms are explored. Precisely, we study the implications of multi-look phase time incongruences on the generation of ground displacement time-series through SB Mt-InSAR methods. Our research clarifies how these phase inconsistencies can propagate through a time-redundant network of SB interferograms and contribute, along with PhU errors, to the quality of the generated ground displacement products. Moreover, we analyze the effects of short-lived phase bias signals that could happen in sequences of short baseline interferograms and propose a strategy for their mitigation. The developed methods have been tested using both simulated and real SAR data. The latter were collected by the Sentinel-1A/B (C-band) sensors over the study areas of Nevada state, U.S., and Sicily Island, Italy. After the development of algorithms for the satellite part, the work veers to ground-based SAR (GB-SAR) sensors. In this field, we propose a method for estimating and compensating the atmospheric phase screen (APS) in sets of SAR interferograms generated with a GB-SAR instrument. We address the presented approach’s physical, statistical, and mathematical framework by discussing its potential and limitations. In contrast with other existing algorithms that estimate the APS from the unwrapped phase signals, our methodology is based on the straightforward analysis of the wrapped phases, directly. Therefore, the method is not affected by any potential phase unwrapping mistake, and it is suitable for Mt-InSAR applications. The effects of the local topography, the decorrelation noise, and the ground deformation on the APS estimates are deeply studied. Experiments performed on simulated and real GB-SAR InSAR data corroborate the validity of the theory. In particular, the simulated results show that the method is beneficial in zones with medium-to-high topographic slopes (e.g., for Alpine and mountainous regions). Further, an interferometric SAR application for the study of three-dimensional (3-D) deformation through the joint and integrated use of satellite and ground SAR data is presented. More precisely, the interferometric data-combining technique exploits the innovative Mt-InSAR algorithms mentioned above, and allows obtaining 3-D mean displacement velocity maps at the finest spatial grid among the available data. In conclusion, also some interested satellite SAR applications in prevention and analysis of particular natural and human-induced disasters are given.

This thesis presents the development of a high-resolution ground-based 3D-SAR system and investigates its application to microwave-vegetation studies. The development process of the system is detailed including an enumeration of high-level requirements, discussions on key design issues, and detailed descriptions of the system down to a component level. The system operates on a 5.4 GHz (C-band) signal, provides a synthetic aperture area of 1.7 m x 1.7 m, and offers resolution of 0.75 m x 0.3 m x 0.3 m (range x azimuth x elevation). The system is employed on several trees with varying physical characteristics. The resulting imagery demonstrates successful 3D reconstruction of the trees and some of their internal features. The individual leaves and small branches are not visible due to the system resolution and the size of the wavelength. The foliage's outline and internal density distribution is resolved. Large branches are visible where geometry is favorable. Trunks are always visible due to their size and normal-facing incidence surface and their return has the strongest contribution from their base. The imagery is analyzed for dependencies on radar and tree parameters including: incidence angle, signal frequency, polarization, inclusion size, water content, and species. In the current work, a single frequency (5.4 GHz) and polarization (HH) is used which leaves the door open for future analysis to use other frequencies and polarizations. The improved resolution capabilities of the 3D-SAR system enables more precise backscatter measurements leading to a greater understanding of microwave-vegetation scattering behavior.

This thesis investigates Ground Moving Target Indication (GMTI) using a multi-aperture Synthetic Aperture Radar (SAR). GMTI relevant methods to process the SAR data collected from a strip-map system are presented. Techniques for mitigation of processing ambiguities are proposed, and a map-drift based automatic focus algorithm for the moving targets is developed. Target detectors based on the multi-channel covariance matrix are presented. Comparisons of the proposed detectors with classical methods, such as Displaced Phase Centre Antenna (DPCA), or Along Track Interferometry (ATI), show increased capability in non-ideal terrain types. By decomposing the complex Wishart probability distribution of the covariance matrix into probability distributions of the eigenvalues and eigenvectors, we derive the receiver operating characteristics (ROCs) of the new detection metrics in ideal terrain. The complete analysis is presented for a two-channel system as well as for special cases of the three-channel system. Finally, since implementation of the Constant False Alarm Rate (CFAR) detectors using the new metrics requires estimation of the local scene parameters, a new parameter estimation method is presented. This parameter estimation technique suggests an extension of the current statistical model of the SAR data; an extension that is also examined.

Synthetic aperture radar (SAR) was first invented in the early 1950s as the remote surveillance instruments to produce high resolution 2D images of the illuminated scene with weather-independent, day-or-night performance. Compared to the Real Aperture Radar (RAR), SAR is synthesising a large virtual aperture by moving a small antenna along the platform path. Typical SAR imaging systems are designed with the basic assumption of a static scene, and moving targets are widely known to induce displacements and defocusing in the formed images. While the capabilities of detection, states estimation and imaging for moving targets with SAR are highly desired in both civilian and military applications, the Ground Moving Target Indication (GMTI) techniques can be integrated into SAR systems to realise these challenging missions. The state-of-the- art SAR-based GMTI is often associated with multi-channel systems to improve the detection capabilities compared to the single-channel ones. Motivated by the fact that the SAR imaging is essentially solving an optimisation problem, we investigate the practicality to reformulate the GMTI process into the optimisation form. Furthermore, the moving target sparsities and underlying similarities between the conventional GMTI processing and sparse reconstruction algorithms drive us to consider the compressed sensing theory in SAR/GMTI applications. This thesis aims to establish an end-to-end SAR/GMTI processing framework regularised by target sparsities based on multi-channel SAR models. We have explained the mathematical model of the SAR system and its key properties in details. The common GMTI mechanism and basics of the compressed sensing theory are also introduced in this thesis. The practical implementation of the proposed framework is provided in this work. The developed model is capable of realising various SAR/GMTI tasks including SAR image formation, moving target detection, target state estimation and moving target imaging. We also consider two essential components, i.e. the data pre-processing and elevation map, in this work. The effectiveness of the proposed framework is demonstrated through both simulations and real data. Given that our focus in this thesis is on the development of a complete sparsity-aided SAR/GMTI framework, the contributions of this thesis can be summarised as follows. First, the effects of SAR channel balancing techniques and elevation information in SAR/GMTI applications are analysed in details. We have adapted these essential components to the developed framework for data pre-processing, system specification estimation and better SAR/GMTI accuracies. Although the purpose is on enhancing the proposed sparsity-based SAR/GMTI framework, the exploitation of the DEM in other SAR/GMTI algorithms may be of independent interest. Secondly, we have designed a novel sparsity-aided framework which integrates the SAR/GMTI missions, i.e. SAR imaging, moving target and background decomposition, and target state estimation, into optimisation problems. A practical implementation of the proposed framework with a two stage process and theoretically/experimentally proven algorithms are proposed in this work. The key novelty on utilising optimisations and target sparsities is explained in details. Finally, a practical algorithm for moving target imaging and state estimation is developed to accurately estimate the full target parameters and form target images with relocation and refocusing capabilities. Compared to the previous processing steps for practical applications, the designed algorithm consistently relies on the exploitation of target sparsities which forms the final processing stage of the whole pipeline. All the developed components contribute coherently to establish a complete sparsity driven SAR/GMTI processing framework.

The use of radar often conjures up images of small blobs on a screen. But current synthetic aperture radar (SAR) systems are able to generate near-optical quality images with amazing benefits compared to optical sensors. These SAR sensors work in all weather conditions, day or night, and provide many advanced capabilities to detect and identify targets of interest. These amazing abilities have made SAR sensors a work-horse in remote sensing, and military applications. SAR sensors are ranging instruments that operate in a 3D environment, but unfortunately the results and interpretation of SAR images have traditionally been done in 2D. Three-dimensional SAR images could provide improved target detection and identification along with improved scene interpretability. As technology has increased, particularly regarding our ability to solve difficult optimization problems, the 3D SAR reconstruction problem has gathered more interest. This dissertation provides the SAR and mathematical background required to pose a SAR 3D reconstruction problem. The problem is posed in a way that allows prior knowledge about the target of interest to be integrated into the optimization problem when known. The developed model is demonstrated on simulated data initially in order to illustrate critical concepts in the development. Then once comprehension is achieved the processing is applied to actual SAR data. The 3D results are contrasted against the current gold- standard. The results are shown as 3D images demonstrating the improvement regarding scene interpretability that this approach provides.

Synthetic aperture radar (SAR) is a type of remote sensor that provides its own illumination and is capable of forming high resolution images of the reflectivity of a scene. The reflectivity of the scene that is measured is dependent on the choice of carrier frequency; different carrier frequencies will yield different images of the same scene. There are different modes for SAR sensors; two common modes are spotlight mode and stripmap mode. Furthermore, SAR sensors can either be continuously transmitting a signal, or they can transmit a pulse at some pulse repetition frequency (PRF). The work in this dissertation is for pulsed stripmap SAR sensors. The resolvable limit of closely spaced reflectors in range is determined by the bandwidth of the transmitted signal and the resolvable limit in azimuth is determined by the bandwidth of the induced azimuth signal, which is strongly dependent on the length of the physical antenna on the SAR sensor. The point-spread function (PSF) of a SAR system is determined by these resolvable limits and is limited by the physical attributes of the SAR sensor. The PSF of a SAR system can be defined in different ways. For example, it can be defined in terms of the SAR system including the image processing algorithm. By using this definition, the PSF is an algorithm-specific sinc-like function and produces the bright, star-like artifacts that are noticeable around strong reflectors in the focused image. The PSF can also be defined in terms of just the SAR system before any image processing algorithm is applied. This second definition of the PSF will be used in this dissertation. Using this definition, the bright, algorithm-specific, star-like artifacts will be denoted as the inter-pixel interference (IPI) of the algorithm. To be specific, the combined effect of the second definition of PSF and the algorithm-dependent IPI is a decomposition of the first definition of PSF. A new comprehensive forward model for stripmap SAR is derived in this dissertation. New image formation methods are derived in this dissertation that invert this forward model and it is shown that the IPI that corrupts traditionally processed stripmap SAR images can be removed. The removal of the IPI can increase the resolvability to the resolution limit, thus making image analysis much easier. SAR data is inherently corrupted by uncompensated phase errors. These phase errors lower the contrast of the image and corrupt the azimuth processing which inhibits proper focusing (to the point of the reconstructed image being unusable). If these phase errors are not compensated for, the images formed by system inversion are useless, as well. A model-based autofocus method is also derived in this dissertation that complements the forward model and corrects these phase errors before system inversion.

Synthetic aperture radar (SAR) is a remote sensing technology for imaging areas of the earth's surface. SAR has been successfully used for monitoring characteristics of the natural environment such as land cover type and tree density. With the advent of higher resolution sensors, it is now theoretically possible to extract information about individual structures such as buildings from SAR imagery. This information could be used for disaster response and security-related intelligence. SAR has an advantage over other remote sensing technologies for these applications because SAR data can be collected during the night and in rainy or cloudy conditions. This research presents a model-based method for extracting information about a building -- its height and roof slope -- from a single SAR image. Other methods require multiple images or ancillary data from specialized sensors, making them less practical. The model-based method uses simulation to match a hypothesized building to an observed SAR image. The degree to which a simulation matches the observed data is measured by mutual information. The success of this method depends on the accuracy of the simulation and on the reliability of the mutual information similarity measure. Electromagnetic theory was applied to relate a building's physical characteristics to the features present in a SAR image. This understanding was used to quantify the precision of building information contained in SAR data, and to identify the inputs needed for accurate simulation. A new SAR simulation technique was developed to meet the accuracy and efficiency requirements of model-based information extraction. Mutual information, a concept from information theory, has become a standard for measuring the similarity between medical images. Its performance in the context of matching a simulation image to a SAR image was evaluated in this research, and it was found to perform well under certain conditions. The factors that affect its performance, and the model-based method overall, were found to include the size of the building and its orientation. Further refinements that expand the range of operational conditions for the method would lead to a practical tool for collecting information about buildings using SAR technology. This research was performed using SAR data from MIT-Lincoln Laboratory.

Radar for biomass estimation has been widely investigated for temperate, boreal and tropical forests, yet tropical savanna woodlands, which generally form non-continuous cover canopies or sparse woodlands, have been largely neglected in biomass studies. This thesis evaluates the capability of Synthetic Aperture Radar (SAR) for estimating the above-ground biomass of the woody vegetation in a savanna in Belize, Central America. This is achieved by evaluating (i) polarimetric Synthetic Aperture Radar (SAR) backscatter and (ii) single-pass shortwave interferometric SAR (InSAR) as indicators of above-ground biomass. Specifically, the effect on SAR backscatter of woody vegetation structure such as canopy cover, basal area, vegetation height and above-ground biomass is evaluated. Since vegetation height is often correlated to above-ground biomass, the effectiveness of vegetation height retrieval from InSAR is evaluated as an indicator of above-ground biomass. The study area, situated in Belize, is representative of Central American savannas. Radar data used are AIRSAR fully polarimetric L- and P-band SAR, and AIRSAR C-band InSAR, Intermap Technologies STAR-3i X-band InSAR, and Shuttle Radar Topography Mission (SRTM) C-band InSAR. The field data comprise accurately georeferenced three-dimensional measurements for 1,133 trees and shrubs and 75 palmetto clumps and thickets in a transect of 800 m x 60 m which spans the main savanna vegetation strata of the study area. An additional 2,464 ground points were observed. Results show that savanna woodlands present a challenge for radar remote sensing methods due to the sparse and heterogeneous nature of savanna woodlands. Long-wave SAR backscatter is dominated not only by high biomass areas, but also by areas of leafy palmetto which have low vegetative biomass. Retrieved woodland canopy heights from X- and C-band InSAR are indicative of the general patterns of tree height, although retrieved heights are underestimated. The amount of underestimation is variable across the different canopy conditions. Of these two methods, the shortwave InSAR data give a better indication of the spatial distribution of the above-ground biomass of the woody vegetation in the savannas than SAR backscatter. These results have implications for new and planned future global biomass estimation missions, such as ALOS PALSAR, ESA’s planned P-band BIOMASS and TanDEM-X. Without appropriate mediation, SAR backscatter methods might overestimate above-ground biomass of the woody vegetation of savannas while InSAR height retrieval methods might underestimate biomass estimates. Some possible mediating approaches are discussed.

International Telemetering Conference Proceedings / October 22-25, 2001 / Riviera Hotel and Convention Center, Las Vegas, Nevada
Synthetic-aperture microwave imaging with ground penetrating radar systems has become a research topic of great importance for the potential applications in sensing and profiling of civil and geophysical structures. It allows us to visualize subsurface structures for nondestructive evaluation with microwave tomographic images. This paper provides an overview of the research program, ranging from the formation of the concepts, physical and mathematical modeling, formulation and development of the image reconstruction algorithms, laboratory experiments, and full-scale field tests.

Synthetic Aperture Radar (SAR) is an imaging technique that creates two dimensional images of the scattering objects in the illuminated ground scene. The objects in the illuminated ground scene may be truly stationary, e.g. buildings etc. or in motion relative to these stationary objects, e.g. cars on a highway. In SAR, the radar platform is moving during the imaging period, hence everything that the radar illuminates has motion relative to the radar platform. In order to specifically detect objects on the ground that are moving relative to stationary ground objects (often termed clutter), processing techniques called Ground Moving Target Indication (GMTI) techniques are required. This is especially required for targets that are moving at relative velocities lower than the stationary clutter's relative velocity to the radar platform (endo-clutter detection). This dissertation investigates five multichannel GMTI techniques being Displaced Phase Centre Antenna (DPCA), Along Track Interferometry (ATI), Iterative Adaptive Approach (IAA), Space Time Adaptive Processing (STAP) and Velocity SAR (VSAR) in literature and assesses the performance of two selected GMTI techniques (ATI and DPCA) on simulated and measured radar data to compare them and identify their strengths and weaknesses. The radar data were measured with a C-band FMCW radar in a controlled environment with known parameters and cooperating targets. The performances of the techniques were assessed in terms of moving target detection within clutter and sensitivity to inaccuracies in the physical system setup. The DPCA technique exhibited some attractive characteristics over the ATI technique. These included its robustness against false alarm in noise dominated cells - ATI exhibited large phase residuals in noise dominated cells, due to the random nature of the phase in these cells. Furthermore, DPCA seem to not suffer from false alarms due to volumetric scattering of vegetation to the extent that was observed with ATI. Lastly, DPCA exhibited more robustness against temporal misalignment errors introduced between the measurement channels, compared to ATI. These observations lead to the conclusion that DPCA would be a practically better choice to implement for the purpose of moving target detection, compared to ATI. However, a double threshold approach, which used DPCA as a pre-processing step to ATI, proved to be superior to DPCA alone in terms of moving target indication within clutter and noise. This approach was verified through implementation on the measured radar data in this study.

The ASCE confers an overall D+ grade to American infrastructure, while the NAE lists the restoration and improvement of urban infrastructure as one of its grand engineering challenges for the 21st century, indicating that infrastructure renovation and development is a major challenge in the US. Furthermore, according to the UN World Urbanization Prospects, about 55% of the world's population lives in urban areas and this percentage is set to grow, especially in Africa and Asia. The growth of urban population poses challenges to the expansion of underground infrastructure, such as water, sewage, electricity and telecommunications. Localization and mapping of underground infrastructure are fundamental for infrastructure maintenance and development. Ground penetrating radar (GPR) is a remote sensing method capable of detecting subsurface assets that has been used in the localization and mapping of underground utilities. This thesis contributes improvements of GPR systems and imaging algorithms towards smarter infrastructure, specifically: Application of GPR imaging algorithm to improve GPR data readability and generate augmented reality (AR) content; Use of photogrammetric methods to improve GPR positioning for underground infrastructure localization and mapping.

Quantifying above-ground biomass (AGB) and carbon sequestration has been a significant focus of attention within the UNFCCC and Kyoto Protocol for improvement of national carbon accounting systems (IPCC, 2007; UNFCCC, 2011). A multitude of research has been carried out in relatively flat and homogeneous forests (Ranson & Sun, 1994; Beaudoin et al.,1994; Kurvonen et al., 1999; Austin et al., 2003; Dimitris et al., 2005), yet forests in the highlands, which generally form heterogeneous forest cover and sparse woodlands with mountainous terrain have been largely neglected in AGB studies (Cloude et al., 2001; 2008; Lumsdon et al., 2005; 2008; Erxue et al., 2009, Tan et al., 2010; 2011a; 2011b; 2011c; 2011d). Since mountain forests constitute approximately 28% of the total global forest area (Price and Butt, 2000), a better understanding of the slope effects is of primary importance in AGB estimation. The main objective of this research is to estimate AGB in the aforementioned forest in Glen Affric, Scotland using both SAR and LiDAR data. Two types of Synthetic Aperture Radar (SAR) data were used in this research: TerraSAR-X, operating at X-band and ALOS PALSAR, operating at L-band, both are fully polarimetric. The former data was acquired on 13 April 2010 and of the latter, two scenes were acquired on 17 April 2007 and 08 June 2009. Airborne LiDAR data were acquired on 09 June 2007. Two field measurement campaigns were carried out, one of which was done from winter 2006 to spring 2007 where physical parameters of trees in 170 circular plots were measured by the Forestry Commission team. Another intensive fieldwork was organised by myself with the help of my fellow colleagues and it comprised of tree measurement in two transects of 200m x 50m at a relatively flat and dense plantation forest and 400m x 50m at hilly and sparse semi-natural forest. AGB is estimated for both the transects to investigate the effectiveness of the proposed method at plot-level. This thesis evaluates the capability of polarimetric Synthetic Aperture Radar data for AGB estimation by investigating the relationship between the SAR backscattering coefficient and AGB and also the relationship between the decomposed scattering mechanisms and AGB. Due to the terrain and heterogeneous nature of the forests, the result from the backscatter-AGB analysis show that these forests present a challenge for simple AGB estimation. As an alternative, polarimetric techniques were applied to the problem by decomposing the backscattering information into scattering mechanisms based on the approach by Yamaguchi (2005; 2006), which are then regressed to the field measured AGB. Of the two data sets, ALOS PALSAR demonstrates a better estimation capacity for AGB estimation than TerraSAR-X. The AGB estimated results from SAR data are compared with AGB derived from LiDAR data. Since tree height is often correlated with AGB (Onge et al., 2008; Gang et al., 2010), the effectiveness of the tree height retrieval from LiDAR is evaluated as an indicator of AGB. Tree delineation was performed before AGB of individual trees were calculated allometrically. Results were validated by comparison to the fieldwork data. The amount of overestimation varies across the different canopy conditions. These results give some indication of when to use LiDAR or SAR to retrieve forest AGB. LiDAR is able to estimate AGB with good accuracy and the R2 value obtained is 0.97 with RMSE of 14.81 ton/ha. The R2 and RMSE obtained for TerraSAR-X are 0.41 and 28.5 ton/ha, respectively while for ALOS PALSAR data are 0.70 and 23.6 ton/ha, respectively. While airborne LiDAR data with very accurate height measurement and consequent three-dimensional (3D) stand profiles which allows investigation into the relationship between height, number density and AGB, it's limited to small coverage area, or large areas but at large cost. ALOS PALSAR, on the other hand, can cover big coverage area but it provide a lower resolution, hence, lower estimation accuracy.

Synthetic aperture radar (SAR) was originally designed as an airborne ground-imaging radar technology. But it has long been desired to also be able to use SAR imaging systems to detect, locate, and track moving ground targets, a process called Ground Moving Target Indication (GMTI). Unfortunately, due to the nature of how SAR works, it is inherently poorly suited to the task of GMTI. SAR only focuses targets and image features that remain stationary during the data collection. A moving ground target therefore does not focus in a conventional SAR image, which complicates the process of performing GMTI with SAR systems. This thesis investigates the feasibility of performing GMTI with single-channel, unsquinted, broadside stripmap SAR despite this inherent limitation. This study focuses solely on the idealized case of direct energy returns from point targets on flat ground, where they and the airborne radar platform all move rectilinearly with constant speed. First, the various aspects of how SAR works, the signal processing used to collect the SAR data, and the backprojection image formation algorithm are explained. The effects of target motion are described and illustrated in actual and simulated SAR images. It is shown how the backprojection (BPJ) algorithm, typically used to image a stationary landscape scene, can also focus on moving targets when the target motion is known a priori. A SAR BPJ ambiguity function is also derived and presented. Next, the time-changing geometry between the airborne radar and a ground target is mathematically analyzed, and it is shown that the slant range between the radar and any ground target, moving or stationary, is a hyperbolic function of time. It is then shown that this hyperbolic range history causes the single-channel SAR GMTI problem to be underdetermined. Finally, a method is then presented for resolving the underdetermined nature of the problem. This is done by constraining a target's GMTI solution using contextual information in the SAR image. Using constraining information, a theoretical way is presented to perform limited GMTI with a single-channel SAR system by using a modified form of the BPJ imaging algorithm, and practical considerations are addressed that complicate the process. Instead of focusing on stationary pixels, this GMTI method uses the BPJ ambiguity function to search for moving targets on a straight path, such as a road, by performing matched filtering on a collection of moving pixels in a position-velocity image space. Nevertheless, it is concluded that for moving point targets, general GMTI with no path constraints is infeasible in practice with a single-channel SAR.

Synthetic Aperture Radar (SAR) is an active microwave remote sensing system able to acquire high resolution images of the scattering behaviour of an observed scene. The contribution of SAR polarimetry (POLSAR) in detection and classification of objects is described and found to add valuable information compared to previous approaches. In this thesis, a new target detection/classification methodology is developed that makes novel use of the polarimetric information of the backscattered field from a target. The detector is based on a geometrical perturbation filter which correlates the target of interest with its perturbed version. Specifically, the operation is accomplished with a polarimetric coherence representing a weighted and normalised inner product between the target and its perturbed version, where the weights are extracted from the observables. The mathematical formulation is general and can be applied to any deterministic (point) target. However, in this thesis the detection is primarily focused on multiple reflections and oriented dipoles due to their extensive availability in common scenarios. An extensive validation against real data is provided exploiting different datasets. They include one airborne system: E-SAR L-band (DLR, German Aerospace Centre); and three satellite systems: ALOS-PALSAR L-band (JAXA, Japanese Aerospace Exploration Agency), RADARSAT-2 C-band (Canadian Space Agency) and TerraSAR-X X-band (DLR). The attained detection masks reveal significant agreement with the expected results based on the theoretical description. Additionally, a comparison with another widely used detector, the Polarimetric Whitening Filter (PWF) is presented. The methodology proposed in this thesis appears to outperform the PWF in two significant ways: 1) the detector is based on the polarimetric information rather than the amplitude of the return, hence the detection is not restricted to bright targets; 2) the algorithm is able to discriminate among the detected targets (i.e. target recognition).

Thesis (M.S.)--University of Missouri-Columbia, 2008.
The entire dissertation/thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file (which also appears in the research.pdf); a non-technical general description, or public abstract, appears in the public.pdf file. Title from title screen of research.pdf file (viewed on July 8, 2009) Includes bibliographical references.

Sparse military satellite constellations were designed using two methods: a traditional approach and a genetic algorithm. One of the traditional constellation designs was the Discoverer II space based radar. Discoverer II was an 8 plane, 24 satellite, Low Earth Orbit (LEO), Walker constellation designed to provide high-range resolution ground moving target indication (HRR-GMTI), synthetic aperture radar (SAR) imaging and high resolution digital terrain mapping. The traditional method designed 9-ball, 12-ball, 18-ball, and 24- ball Walker constellations. The genetic algorithm created constellations by deriving a phenotype from a triploid genotype encoding of orbital elements. The performance of both design methods were compared using a computer simulation. The fitness of each constellation was calculated using maximum gap time, maximum revisit time, and percent coverage. The goal was to determine if one design method would consistently outperform the other. The genetic algorithm offered a fitness improvement over traditional constellation design methods in all cases except the 24-ball constellation where it demonstrated comparable results. The genetic algorithm improvement over the traditional constellations increased as the number of satellites per constellation decreased. A derived equation related revisit time to the number of ship tracks maintained.
US Navy (USN) author.

International Telemetering Conference Proceedings / October 20-23, 2003 / Riviera Hotel and Convention Center, Las Vegas, Nevada
In this paper, the research prototype of a high-performance GPR imaging system is presented. The system is equipped with the capability of synthetic-aperture scan, stepfrequency FMCW illumination, and high-resolution tomographic image reconstruction.

The capability to observe ocean swell using spaceborne Synthetic Aperture Radar (SAR) has been demonstrated starting with ERS-1 mission in 1992. This dissertation shows how ocean swell properties can be used to combine swell observations of heterogeneous quality and acquired at various times and locations for the observation and forecast of ocean swell fieldsusing ASAR instrument on-board ENVISAT. The first section is a review of how ocean swell spectra can be derived from the SAR complex images of the ocean surface using a quasi-linear transformation. Then, significant swell heights, peak periods and peak directions from in situ measurements are used to assess the accuracy of the SAR observed swell spectra. Using linear propagation in deep ocean, a new swell field reconstruction methodologyis developed in order to gather SAR swell observations related to the same swell field. Propagated from their generation region, these observations render the spatio-temporal properties of the emanating ocean swell fields. Afterwards, a methodology is developed for the exclusion of outliers taking advantage of the swell field consistency. Also, using the irregularly sampled SAR observations, quality controlled estimations of swell field integral parameters are produced on a regular space-time grid. Validation against in situ measurements reveals the dramatic impact of the density of propagated observations on the integral parameters estimated accuracy. Specifically, this parameter is shown to be very dependent on the satellite orbit. Finally, comparisons with the numerical wave model WAVEWATCH-III prove it could potentially benefit from the SAR swell field estimates for assimilation purposes.

One problem of interest to the oceanic engineering community is the detection and enhancement of internal wakes in open water synthetic aperture radar (SAR) images. Internal wakes, which occur when a ship travels in a stratified medium, have a "V" shape extending from the ship, and a chirp-like feature across each arm. The Radon transform has been applied to the detection and the enhancement problems in internal wake images to account for the linear features while the wavelet transform has been applied to the enhancement problem in internal wake images to account for the chirp-like features. Although the Radon transform accentuates linear features, there have been several difficulties applying this transform to the wake detection and enhancement problem because the transform is not localized. In a recent article by Copeland et. al., a localized Radon transform (LRT) was developed and was shown to reduce the speckle noise. In this dissertation, another derivation of the LRT is obtained which shows that this transform is equivalent to the Radon transform with a rectangular window function. Several properties not considered in the article are derived using the new formulation. Another transform which has been applied to internal wake images is the wavelet transform. In a recent paper by Teti et. al., the wavelet transform was applied to slices through internal wakes in SAR images. Although the wavelet transform reduced the speckle noise in SAR wake images, it required extracting a line from the image. In this dissertation, a wavelet localized Radon transform is developed which performs the wavelet transform on all lines in an image without explicitly extracting slices of the image. The fundamental theory for this transform is developed and several examples are considered. This transform is then expanded to include features which occur over a region with a significant length. The fundamental theory for this new transform, a localized Radon transform with a wavelet filter, is developed and several examples are provided. These new transforms are then incorporated into optimal and sub-optimal detection schemes for images with linear features, including ship wakes, which are contaminated by additive Gaussian noise.

Land cover classification using Synthetic Aperture Radar (SAR) data has been a topic of great interest in recent literature. Food commodities output prediction through crop identification, environmental monitoring, and forest regrowth tracking are some of the many problems that can be aided by land cover classification methods. The need for fast and automated classification methods is apparent in a variety of applications involving vast amounts of SAR data. One fundamental step in any supervised learning classification algorithm is the selection and/or extraction of features present in the dataset to be used for class discrimination. A popular method that has been proposed for feature extraction from polarimetric data is to decompose the data into the underlying scattering mechanisms. In this research, the Freeman and Durden scattering model is applied to ALOS PALSAR fully polarimetric data for feature extraction. Efficient methods for solving the complex system of equations present in the scattering model are developed and compared. Using the features from the Freeman and Durden work, the classification capability of the model is assessed on amazon rainforest land cover types using a supervised Support Vector Machine (SVM) classification algorithm. The quantity of land cover types that can be discriminated using the model is also determined. Additionally, the performance of the median as a robust estimator in noisy environments for multi-pixel windowing is also characterized.
Master of Science
Land type classification using Radar data has been a topic of great interest in recent literature. Food commodities output prediction through crop identification, environmental monitoring, and forest regrowth tracking are some of the many problems that can be aided by land cover classification methods. The need for fast and automated classification methods is apparent in a variety of applications involving vast amounts of Radar data. One fundamental step in any classification algorithm is the selection and/or extraction of discriminating features present in the dataset to be used for class discrimination. A popular method that has been proposed for feature extraction from polarized Radar data is to decompose the data into the underlying scatter components. In this research, a scattering model is applied to real world data for feature extraction. Efficient methods for solving the complex system of equations present in the scattering model are developed and compared. Using the features from the scattering model, the classification capability of the model is assessed on amazon rainforest land types using a Support Vector Machine (SVM) classification algorithm. The quantity of land cover types that can be discriminated using the model is also determined and compared using different estimators.

Direct GPS signals are used for navigation and positioning purposes by a diverse set of users. However, this project describes a method for utilizing reflected Global Positioning System (GPS) signals to form an image of targets within a region of interest. The principle is based upon a type of bi-static synthetic aperture radar (SAR) in which a matched filter technique is employed to perform the image reconstruction by the correlation of direct and reflected GPS signals. This method relies upon the fact that each component of the received signal resulting from a reflection of an individual target is subjected to a unique chirp. A major challenge was the appalling signal to noise ratio associated with the received reflected GPS signals. The reconstruction method resulted in an undesirable point spread function (PSF) which seriously smears the reconstructed.

Estimation of biomass has high importance for economic, ecologic and climatic reasons due to the multiple ecosystem services offered by forested landscapes. Measurements that are taken in the field incur personal and economic costs. Nevertheless, biomass surveying based on remote sensing techniques offer efficiency thanks to covering large areas. The European Space Agency (ESA) Sentinel-1 satellite offers promising capabilities for above-ground biomass (AGB) estimation through synthetic aperture radar (SAR) based microwave remote sensing. In this study, experimental AGB estimations based on Sentinel-1 C-band data were produced over the Remingstorp estate (Västergötland County, Sweden) to analyze its performance over boreal productive forests. The obtained measurements were compared against reference values obtained by combining photogrammetric, aerial laser scanning (ALS) and field measurements. Thus, a reference high-resolution canopy height model (CHM) was produced from the difference between photogrammetric digital surface model (DSM) values and ALS digital terrain model (DTM) values. The comparison of CHM observations against diameter at breast height (DBH) field measurements revealed the existence of a vegetation height - vegetation volume relationship for the study species (Pinus Sylvestris and Picea Abbies), which allowed bole volume estimation based on vegetation height values. SAR-based AGB estimates were produced by defining statistical relationships between backscatter intensity and interferometric coherence measurements against reference CHM values. Additionally, evaluation of biomass estimation through interferometric (InSAR) height was possible by comparing against reference photogrammetric DSM. Backscatter signal saturation of C-band at low biomass volumes prevented quantification of biomass but permitted differentiation between forested and non-forested surfaces. Estimation of AGB through interferometric coherence was possible through modeling volumetric decorrelation, which on the contrary prevented biomass retrieval from InSAR height. Due to the given frequency properties at C-band, HV cross-polarized channel was used in all cases for better detection of the canopy layer. Image acquisition under stable conditions was a priority to avoid noise derived from variable dielectric properties, acquisition geometry effects and temporal decorrelation. Hence, image acquisitions under stable hydrometeorological conditions (i. e. stable frozen or dry) and for the lowest repeat-pass interval (i. e. 6-days) were prioritized.

Thesis (MA)--Stellenbosch University, 2015.
ENGLISH ABSTRACT: Image classification has long been used in earth observation and is driven by the need for accurate maps to develop conceptual and predictive models of Earth system processes. Synthetic aperture radar (SAR) imagery is used ever more frequently in land cover classification due to its complementary nature with optical data. There is therefore a growing need for reliable, accurate methods for using SAR and optical data together in land use and land cover classifications. However, combining data sets inevitably increases data dimensionality and these large, complex data sets are difficult to handle. It is therefore important to assess the benefits and limitations of using multi-temporal, dual-sensor data for applications such as land cover classification. This thesis undertakes this assessment through four main experiments based on combined RADARSAT-2 and SPOT-5 imagery of the southern part of Reunion Island. In Experiment 1, the use of feature selection for dimensionality reduction was considered. The rankings of important features for both single-sensor and dual-sensor data were assessed for four dates spanning a 6-month period, which coincided with both the wet and dry season. The mean textural features produced from the optical bands were consistently ranked highly across all dates. In the two later dates (29 May and 9 August 2014), the SAR features were more prevalent, showing that SAR and optical data have complementary natures. SAR data can be used to separate classes when optical imagery is insufficient. Experiment 2 compared the accuracy of six supervised and machine learning classification algorithms to determine which performed best with this complex data set. The Random Forest classification algorithm produced the highest accuracies and was therefore used in Experiments 3 and 4. Experiment 3 assessed the benefits of using combined SAR-optical imagery over single-sensor imagery for land cover classifications on four separate dates. The fused imagery produced consistently higher overall accuracies. The 29 May 2014 fused data produced the best accuracy of 69.8%. The fused classifications had more consistent results over the four dates than the single-sensor imagery, which suffered lower accuracies, especially for imagery acquired later in the season. In Experiment 4, the use of multi-temporal, dual-sensor data for classification was evaluated. Feature selection was used to reduce the data set from 638 potential training features to 50, which produced the best accuracy of 74.1% in comparison to 71.9% using all of the features. This result validated the use of multi-temporal data over single-date data for land cover classifications. It also validated the use of feature selection to successfully inform data reduction without compromising the accuracy of the final product. Multi-temporal and dual-sensor data shows potential for mapping land cover in a tropical, mountainous region that would otherwise be challenging to map using single-sensor data. However, accuracies Stellenbosch University https://scholar.sun.ac.za iv generally remained lower than would allow for transferability and replication of the current methodology. Classification algorithm optimisation, supervised segmentation and improved training data should be considered to improve these results.
AFRIKAANSE OPSOMMING: Beeld-klassifikasie word al ‘n geruime tyd in aardwaarneming gebruik en word gedryf deur die behoefte aan akkurate kaarte om konseptuele en voorspellende modelle van aard-stelsel prosesse te ontwikkel. Sintetiese apertuur radar (SAR) beelde word ook meer dikwels in landdekking klassifikasie gebruik as gevolg van die aanvullende waarde daarvan met optiese data. Daar is dus 'n groeiende behoefte aan betroubare, akkurate metodes vir die gesamentlike gebruik van SAR en optiese data in landdekking klassifikasies. Die kombinasie van datastelle bring egter ‘n onvermydelike verhoging in data dimensionaliteit mee, en hierdie groot, komplekse datastelle is moeilik om te hanteer. Dus is dit belangrik om die voordele en beperkings van die gebruik van multi-temporale, dubbel-sensor data vir toepassings soos landdekking-klassifikasie te evalueer. Die waarde van gekombineerde (versmelte) RADARSAT-2 en SPOT-5 beelde word in hierdie tesis deur middel van vier eksperimente geevalueer. In Eksperiment 1 is die gebruik van kenmerk seleksie vir dimensionaliteit-vermindering toegepas. Die ranglys van belangrike kenmerke vir beide enkel-sensor en 'n dubbel-sensor data is beoordeel vir vier datums wat oor 'n tydperk van 6 maande strek. Die gemiddelde tekstuur kenmerke uit die optiese lae is konsekwent hoog oor alle datums geplaas. In die twee later datums (29 Mei en 9 Augustus 2014) was die SAR kenmerke meer algemeen, wat dui op die aanvullende aard van SAR en optiese data. SAR data dus gebruik kan word om klasse te onderskei wanneer optiese beelde onvoldoende daarvoor is. Eksperiment 2 het die akkuraatheid van ses gerigte en masjien-leer klassifikasie algoritmes vergelyk om te bepaal watter die beste met hierdie komplekse datastel presteer. Die random gorest klassifikasie algoritme het die hoogste akkuraatheid bereik en is dus in Eksperimente 3 en 4 gebruik. Eksperiment 3 het die voordele van gekombineerde SAR-optiese beelde oor enkel-sensor beelde vir landdekking klassifikasies op vier afsonderlike datums beoordeel. Die versmelte beelde het konsekwent hoër algehele akkuraathede as enkel-sensor beelde gelewer. Die 29 Mei 2014 data het die hoogste akkuraatheid van 69,8% bereik. Die versmelte klassifikasies het ook meer konsekwente resultate oor die vier datums gelewer en die enkel-sensor beelde het tot laer akkuraathede gelei, veral vir die later datums. In Eksperiment 4 is die gebruik van multi-temporale, dubbel-sensor data vir klassifikasie ge-evalueer. Kenmerkseleksie is gebruik om die data stel van 638 potensiële kenmerke na 50 te verminder, wat die beste akkuraatheid van 74,1% gelewer het. Hierdie resultaat bevestig die belangrikheid van multi-temporale data vir grond dekking klassifikasies. Dit bekragtig ook die gebruik van kenmerkseleksie om data vermindering suksesvol te rig sonder om die akkuraatheid van die finale produk te belemmer. Stellenbosch University https://scholar.sun.ac.za vi Multi-temporale en dubbel-sensor data toon potensiaal vir die kartering van landdekking in 'n tropiese, bergagtige streek wat andersins uitdagend sou wees om te karteer met behulp van enkel-sensor data. Oor die algemeen het akkuraathede egter te laag gebly om vir oordraagbaarheid en herhaling van die huidige metode toe te laat. Klassifikasie algoritme optimalisering, gerigte segmentering en verbeterde opleiding data moet oorweeg word om hierdie resultate te verbeter.

Automatic Target Recognition (ATR) is a subject involving the use of sensor data to develop an algorithm for identifying targets of significance. It is of particular interest in military applications such as unmanned aerial vehicles and missile tracking systems. This thesis develops an orientation-based classification approach from previous ATR algorithms for 2-D Synthetic Aperture Radar (SAR) images. Prior work in ATR includes Chessa Guilas’ Hausdorff Probabilistic Feature Analysis Approach in 2005 and Daniel Cary’s Optimal Rectangular Fit in 2007. A system incorporating multiple modules performing different tasks is developed to streamline the data processing of previous algorithms. Using images from the publicly available Moving and Stationary Target Acquisition and Recognition (MSTAR) database, target orientation was determined to be the best feature for ATR. A rotationally variant algorithm taking advantage of the combination of target orientation and pixel location for classification is proposed in this thesis. Extensive classification results yielding an overall accuracy of 76.78% are presented to demonstrate algorithm functionality.

Arctic sea ice is rapidly declining in extent, thickness, volume and age, with the majority of the decline in extent observed at the end of the melt season. Advanced melt is a thermodynamic regime and is characterized by the formation of melt ponds on the sea ice surface, which have a lower surface albedo (0.2-0.4) than the surrounding ice (0.5-0.7) allowing more shortwave radiation to enter the system. The loss of multiyear ice (MYI) may have a profound impact on the energy balance of the system because melt ponds on first-year ice (FYI) comprise up to 70% of the ice surface during advanced melt, compared to 40% on MYI. Despite the importance of advanced melt to the ocean-sea ice-atmosphere system, advanced melt and the extent to which winter conditions influence it remain poorly understood due to the highly dynamic nature of melt pond formation and evolution, and a lack of reliable observations during this time. In order to establish quantitative links between winter and subsequent advanced melt conditions, and assess the effects of scale and choice of aggregation features on the relationships, three data aggregation approaches at varied spatial scales were used to compare high resolution satellite GeoEye-1 optical images of melt pond covered sea ice to winter airborne laser scanner surface roughness and electromagnetic induction sea ice thickness measurements. The findings indicate that winter sea ice thickness has a strong association with melt pond fraction (fp) for FYI and MYI. FYI winter surface roughness is correlated with fp, whereas for MYI no association with fp was found. Satellite-borne synthetic aperture radar (SAR) data are heavily relied upon for sea ice observation; however, during advanced melt the reliability of observations is reduced. In preparation for the upcoming launch of the RADARSAT Constellation Mission (RCM), the Kolmogorov-Smirnov (KS) statistical test was used to assess the ability of simulated RCM parameters and grey level co-occurrence matrix (GLCM) derived texture features to discriminate between major ice types during winter and advanced melt, with a focus on advanced melt. RCM parameters with highest discrimination ability in conjunction with optimal GLCM texture features were used as input parameters for Support Vector Machine (SVM) supervised classifications. The results indicate that steep incidence angle RCM parameters show promise for distinguishing between FYI and MYI during advanced melt with an overall classification accuracy of 77.06%. The addition of GLCM texture parameters improved accuracy to 85.91%. This thesis provides valuable contributions to the growing body of literature on fp parameterization and SAR ice type discrimination during advanced melt.
Graduate
2019-03-21

In Inverse Synthetic Aperture Radar (ISAR) systems the motion of the target can be classified in two main categories: Translational Motion and Rotational Motion. A small degree of rotational motion is required in order to generate the synthetic aperture of the ISAR systems. On the other hand, the remaining part of the target&rsquo
s motion, that is any degree of translational motion and the large degree of rotational motion, degrades ISAR image quality. Motion compensation techniques focus on eliminating the effect of the targets&rsquo
motion on the ISAR images. In this thesis, ISAR image generation is discussed using both Conventional Fourier Based and Time-Frequency Based techniques. Standard translational motion compensation steps, Range and Doppler Tracking, are examined. Cross-correlation method and Dominant Scatterer Algorithm are employed for Range and Doppler tracking purposes, respectively. Finally, Time-Frequency based motion compensation is studied and compared with the conventional techniques. All of the motion compensation steps are examined using the simulated data. Stepped frequency waveforms are used in order to generate the required data of the simulations. Not only successful results, but also worst case examinations and lack of algorithms are also discussed with the examples.

Ground-based/terrestrial radar interferometry (GBRI) is a scientific topic of increasing interest in recent years. The GBRI is used in several field as remote sensing technique for monitoring natural environment (landslides, glacier, and mines) or infrastructures (bridges, towers). These sensors provide the displacement of targets by measuring the phase difference between sending and receiving radar signal. If the acquisition rate is enough the GBRI can provide the natural frequency, e.g. by calculating the Fourier transform of displacement. The research activity, presented in this thesis, concerns design and development of some advanced GBRI systems. These systems are related to the following issue: detection of displacement vector, Multiple Input Multiple Output (MIMO) and radars with 3D capability. The conventional GBRI measures only the component of displacement along range direction. A GBRI operating in monostatic and bistatic modality is presented in this thesis. The sensor detects the first component of displacement as the conventional GBRI (monostatic) and an additional component through a transponder (bistatic). The radar has been successfully tested in controlled environment using a basic transponder (two antennas and an amplifier). The transponder has been improved to increase the gain of the amplifier and to solve some issue of the basic version. Finally, the system is used in real application for measuring the natural axis of a telecommunication tower. The most advanced GRBI system can measure the directional of arrival of scattered signal by exploiting the movement of the antenna on an axis (Ground Based Synthetic Aperture Radar - GBSAR). The step between two position on the axis has to be smaller than a quarter of wavelength. The emerging Multiple Input Multiple Output (MIMO) technique can be used to reduce the mechanical movement parts and the problems related to these. Also, for MIMO radar the spacing between two closer phase center has to be smaller than a quarter of wavelength for the Shannon theorem. In this thesis a Compressive Sensing (CS) MIMO radar is described. Indeed, the CS is a technique able to reconstruct signal without the constrain of Shannon theorem. The signal has to be sparse and randomly sampled in order to use the CS. The CS technique can be applied for increase the scan-length of a MIMO system of $40%div50%$. Therefore, by using the same number of antennas, the CS allows to increase the angular resolution of a MIMO radar. A prototype of interferometric CS MIMO radar has been developed and tested on some bridges. The results were compared with a conventional GBRI with a good agreement. The CS MIMO radar was able to discriminate the left-right movement of bridges. Unfortunately, the repetition rate of this prototype was not enough to retrieve the spectra of natural frequency. Since the movement is along a single axis the obtained radar image does not have angular resolution in the plane orthogonal to the scan axis. In other words, if the radar head scans along the x-axis the radar image cannot have resolution in elevation angle. This is not a serious problem when the scenario is a slope, where the elevation (z-axis) can be reasonably considered an unambiguous function of the (x,y) position. Unfortunately, there are cases where the geometry of the structure under test is much more complex, i.e in urban environment. In this thesis two radar systems with three-dimensional resolution are reported. These two systems synthesize the two technique previously described. Indeed, the first sensor uses the bistatic principle by exploiting the movement of an additional antenna in vertical axis for obtaining the resolution in elevation. The second system exploits the movement on a horizontal axis of the CS MIMO with phase center positioned on a vertical axis. In order to test the capability, the two radars were located in an urban scenario in front of a 7-storey building. Both systems were able to provide a 3D image of the building.

碩士
元智大學
通訊工程學系
104
This thesis is based on space program at Yuan Ze University’s communication center. However, design of satellite-based Synthetic Aperture Radar(SAR) needs high power module and higher cost. So we design RF front-end module for C-band ground-based SAR which is lower band. It follows by SAR principle and we test all RF components for RF front-end transceiver. In addition, we design a module for C-band ground-based SAR combined with signal program which can transmit LFM pulse signal and signalprocessing. In the test, we slide the RF Front-End Module to observe atarget, and use notebook to capture the received signal from the scope stored for further SAR processing. Finally, we will analyze imaging correctness.