High Geometric Resolution Research Articles

ADVANCES IN HYPERSPECTRAL AND MULTISPECTRAL IMAGE FUSION AND SPECTRAL UNMIXING

Abstract. In this work, we jointly process high spectral and high geometric resolution images and exploit their synergies to (a) generate a fused image of high spectral and geometric resolution; and (b) improve (linear) spectral unmixing of hyperspectral endmembers at subpixel level w.r.t. the pixel size of the hyperspectral image. We assume that the two images are radiometrically corrected and geometrically co-registered. The scientific contributions of this work are (a) a simultaneous approach to image fusion and hyperspectral unmixing, (b) enforcing several physically plausible constraints during unmixing that are all well-known, but typically not used in combination, and (c) the use of efficient, state-of-the-art mathematical optimization tools to implement the processing. The results of our joint fusion and unmixing has the potential to enable more accurate and detailed semantic interpretation of objects and their properties in hyperspectral and multispectral images, with applications in environmental mapping, monitoring and change detection. In our experiments, the proposed method always improves the fusion compared to competing methods, reducing RMSE between 4% and 53%.

A Novel Graph-Matching-Based Approach for Domain Adaptation in Classification of Remote Sensing Image Pair

This paper addresses the problem of land-cover classification of remotely sensed image pairs in the context of domain adaptation. The primary assumption of the proposed method is that the training data are available only for one of the images (source domain), whereas for the other image (target domain), no labeled data are available. No assumption is made here on the number and the statistical properties of the land-cover classes that, in turn, may vary from one domain to the other. The only constraint is that at least one land-cover class is shared by the two domains. Under these assumptions, a novel graph theoretic cross-domain cluster mapping algorithm is proposed to detect efficiently the set of land-cover classes which are common to both domains as well as the additional or missing classes in the target domain image. An interdomain graph is introduced, which contains all of the class information of both images, and subsequently, an efficient subgraph-matching algorithm is proposed to highlight the changes between them. The proposed cluster mapping algorithm initially clusters the target domain data into an optimal number of groups given the available source domain training samples. To this end, a method based on information theory and a kernel-based clustering algorithm is proposed. Considering the fact that the spectral signature of land-cover classes may overlap significantly, a postprocessing step is applied to refine the classification map produced by the clustering algorithm. Two multispectral data sets with medium and very high geometrical resolution and one hyperspectral data set are considered to evaluate the robustness of the proposed technique. Two of the data sets consist of multitemporal image pairs, while the remaining one contains images of spatially disjoint geographical areas. The experiments confirm the effectiveness of the proposed framework in different complex scenarios.

The possibility of synthesizing the aperture of a satellite thermal viewer

Satellite thermal-vision monitoring requires both high geometrical resolution and frequent repetition of the imaging. However, high-angular-resolution satellite thermal viewers have a narrow viewing band and do not allow regular monitoring of territories. A resolution of this contradiction is proposed on the basis of the synthesis of the angular aperture of a scanning radiometer (thermal viewer). The results of laboratory and numerical experiments on the synthesis of the angular aperture of a thermal viewer are presented. It is shown that the synthetic angular aperture can be several times as large as the effective angular resolution of the thermal viewer without changing the viewing angle, and that the SNR is acceptable in practice.

MOCO for High-Resolution ScanSAR via Full-Aperture Processing

In this paper, a novel Doppler rate (DR) estimation-based two-dimensional (2-D) subaperture motion compensation (MOCO) approach is proposed for high-resolution scanning synthetic aperture radar (ScanSAR) mode. It can estimate space-variant phase errors in different subswaths simultaneously by exploiting data of multiple subswaths. High-resolution ScanSAR images, produced by full-aperture processing algorithm, are often defocused by uncompensated phase errors. As a result of the periodic gapped data in each subswath in ScanSAR mode, the performance of the commonly used stripmap MOCO approaches will be degraded or even not work when using full-aperture processing algorithm. The effectiveness of the proposed approach is demonstrated by an airborne ScanSAR real data set containing different types of terrain, with a high geometric resolution of about 3.5 m (mid or far range) or 5 m (near range) in azimuth working at C band.

ASTC-MIMO-TOPS Mode with Digital Beam-Forming in Elevation for High-Resolution Wide-Swath Imaging

Future spaceborne synthetic aperture radar (SAR) missions require complete and frequent coverage of the earth with a high resolution. Terrain Observation by Progressive Scans (TOPS) is a novel wide swath mode but has impaired azimuth resolution. In this paper, an innovative extended TOPS mode named Alamouti Space-time Coding multiple-input multiple-output TOPS (ASTC-MIMO-TOPS) mode combined with digital beam-forming (DBF) in elevation and multi-aperture SAR signal reconstruction in azimuth is proposed. This innovative mode achieves wide-swath coverage with a high geometric resolution and also overcomes major drawbacks in conventional MIMO SAR systems. The data processing scheme of this imaging scheme is presented in detail. The designed system example of the proposed ASTC-MIMO-TOPS mode, which has the imaging capacity of a 400 km wide swath with an azimuth resolution of 3 m, is given. Its system performance analysis results and simulated imaging results on point targets demonstrate the potential of the proposed novel spaceborne SAR mode for high-resolution wide-swath (HRWS) imaging.

Derivation of temporal windows for accurate crop discrimination in heterogeneous croplands of Uzbekistan using multitemporal RapidEye images

Recently launched and upcoming satellite systems are expected to make crop mapping increasingly accurate and operational. This study aims at an evaluation of the optimum number of acquisition dates and most suitable temporal windows for the discrimination of crops over heterogeneous agricultural landscapes utilizing multitemporal data from the RapidEye satellite constellation. For a case study site in West-Uzbekistan, which covers 230,000ha of irrigated land, nine 6.5m RapidEye mosaics were obtained throughout the vegetation period of 2009. The acquisition window of one mosaic was maximum five days. The optimum number of acquisition dates and acquisition windows relevant for accurate crop identification was assessed by applying a random forest algorithm to all possible combinations of the available RapidEye mosaics.Overall accuracy increased with the number of acquisition dates selected for the classification. Classifying all nine mosaics resulted in an overall accuracy of 85.7%. This accuracy level could be already achieved by using five or more acquisitions during distinctive phases of the vegetation period, among others short before winter-wheat harvest and in initial growing phases of summer crops. It could be shown that accurate early season crop maps at field level can be derived enabling water managers to update water distribution schedules in the study region. All 53,302 fields of the irrigation system including very small parcels with a size of 0.05ha could be classified due to the high geometric resolution of RapidEye. Detailed NDVI temporal profiles demonstrated the potential of RapidEye for including minor crops into classifications of heterogeneous agricultural areas. Class-wise accuracy pointed to the challenge of classifying sub-tree cultivation and indicated that classification of minor crops could also depend on the spatial resolution of satellite images. The case study suggests identifying temporal windows for crop mapping serving as substantial support for land and water management in different agro-environmental settings.

A framework for the geometric accuracy assessment of classified objects

European initiatives for data harmonization and the establishment of remote-sensing-based services aim at the production of up-to-date land-cover information according to generally valid standards for the accurate qualification of thematic classification results. This is particularly true since new satellite systems provide data of high temporal and geometric resolution. While methods for point-related thematic accuracy assessment have already been established for years, there is a need for a commonly accepted framework for the geometric quality of tematic maps. In this study, an open and extendable framework for the geometric accuracy assessment is presented. The workflow begins with the definition of basic geometric accuracy metrics, which are based on differences in area and position between samples of classified and reference objects. The combination of user-defined metrics enables both a geometric assessment of single objects as well as the total data set. In an example of thematically classified agricultural fields in a German test site, we finally show how object relations between classified and reference objects can be identified and how they affect the global accuracy assessment of the total data set.

Multichannel synthetic aperture radar systems with a planar antenna for future spaceborne microwave remote sensing

Future spaceborne SAR missions will require complete and frequent coverage of the Earth with a reasonably high geometric resolution. To achieve this imaging capacity, multichannel SAR systems with a planar antenna will be widely adopted in the next generation spaceborne microwave remote sensing missions. However, the high data rate, the increased system complexity, the difficulty of multichannel signal processing, the heavy load, and the wide bandwidth downlink may hamper the demonstration of these imaging modes. With the rapid development of electronic engineering and digital signal processing, these problems will be well resolved. Furthermore, regarding the implementation complexity, multichannel SAR systems with a deployable reflector antenna seem to be a better alternative for HRWS imaging [17, 18]. However, the major drawback of the reflector antenna is the limited beam-steering capacity compared with a planar antenna, which will affect some of the remote sensing applications. Thus, multichannel SAR systems with a planar antenna rather than a reflector are more suitable for the higher geometric resolution, especially for the higher-carrier frequency (such as X-band).

Galileo-based space–airborne bistatic SAR for UAS navigation

This paper presents the idea of a hybrid bistatic Synthetic Aperture Radar (SAR) formed by a constellation of spaceborne illuminators and receiving-only airborne SARs. The study aims at identifying and developing new remote sensing techniques, able to produce forward-looking, real-time, any-weather, day and night, high geometric and radiometric resolution data for Unmanned Airborne Systems (UAS) autonomous navigation, namely obstacle detection and landing. Special emphasis is devoted to potentials and limitations of using Galileo or a Galileo improved constellation as illuminator. Preliminary feasibility assessment and definition of strawman concepts conducted under ESA contract in the framework of future integrated aerospace systems are reported, consisting in performance and trade-off studies of two configurations of the space segment and of two operating modes of the space–airborne bistatic SAR.

Multidetector computed tomography of chest trauma: indications, technique and interpretation

BackgroundChest traumas are a significant cause of mortality and morbidity, especially in the younger population.MethodsDiagnostic imaging plays a key role in their management. Multidetector computed tomography (MDCT) is the most important imaging method in this field. Its advantages include especially high speed and high geometric resolution in any plane.ResultsThe method allows us to view large parts of the body with minimal motion artifacts and to create accurate multiplanar and three-dimensional (3D) reformations, which make the diagnosis significantly more accurate. Because of its advantages MDCT has become the first-choice method in high-energy traumas.ConclusionThis article summarises the position of MDCT in the diagnostic algorithm of chest injuries, technical aspects of the examination and imaging findings in traumas of the individual chest compartments.Teaching Points• Diagnostic imaging plays a key role in the management of high-energy chest trauma.• MDCT is the most important imaging method in this kind of injury, as detailed information can be acquired in a short acquisition time.• Multiplanar and three-dimensional (3D) reformattings make the diagnosis significantly more accurate.

ROAD DETECTION BY NEURAL AND GENETIC ALGORITHM IN URBAN ENVIRONMENT

Abstract. In the urban object detection challenge organized by the ISPRS WG III/4 high geometric and radiometric resolution aerial images about Vaihingen/Stuttgart, Germany are distributed. The acquired data set contains optical false color, near infrared images and airborne laserscanning data. The presented research focused exclusively on the optical image, so the elevation information was ignored. The road detection procedure has been built up of two main phases: a segmentation done by neural networks and a compilation made by genetic algorithms. The applied neural networks were support vector machines with radial basis kernel function and self-organizing maps with hexagonal network topology and Euclidean distance function for neighborhood management. The neural techniques have been compared by hyperbox classifier, known from the statistical image classification practice. The compilation of the segmentation is realized by a novel application of the common genetic algorithm and by differential evolution technique. The genes were implemented to detect the road elements by evaluating a special binary fitness function. The results have proven that the evolutional technique can automatically find major road segments.

ANALYSIS OF MOBILE LASER SCANNING DATA AND MULTI-VIEW IMAGE RECONSTRUCTION

Abstract. The combination of laser scanning (LS, active, direct 3D measurement of the object surface) and photogrammetry (high geometric and radiometric resolution) is widely applied for object reconstruction (e.g. architecture, topography, monitoring, archaeology). Usually the results are a coloured point cloud or a textured mesh. The geometry is typically generated from the laser scanning point cloud and the radiometric information is the result of image acquisition. In the last years, next to significant developments in static (terrestrial LS) and kinematic LS (airborne and mobile LS) hardware and software, research in computer vision and photogrammetry lead to advanced automated procedures in image orientation and image matching. These methods allow a highly automated generation of 3D geometry just based on image data. Founded on advanced feature detector techniques (like SIFT (Scale Invariant Feature Transform)) very robust techniques for image orientation were established (cf. Bundler). In a subsequent step, dense multi-view stereo reconstruction algorithms allow the generation of very dense 3D point clouds that represent the scene geometry (cf. Patch-based Multi-View Stereo (PMVS2)). Within this paper the usage of mobile laser scanning (MLS) and simultaneously acquired image data for an advanced integrated scene reconstruction is studied. For the analysis the geometry of a scene is generated by both techniques independently. Then, the paper focuses on the quality assessment of both techniques. This includes a quality analysis of the individual surface models and a comparison of the direct georeferencing of the images using positional and orientation data of the on board GNSS-INS system and the indirect georeferencing of the imagery by automatic image orientation. For the practical evaluation a dataset from an archaeological monument is utilised. Based on the gained knowledge a discussion of the results is provided and a future strategy for the integration of both techniques is proposed.

THE MICROSOFT GLOBAL ORTHO PROGRAM

Abstract. Wide area and thus continental mapping extending beyond national borders is a novel concept in civilian photogrammetry. The Microsoft Global Ortho Program was launched in the Spring of 2009 as a result of Microsoft's need for global geo-data at a high geometric resolution and radiometric excellence. By fall of 2012 more than 10 million km2 of the USA and 14 European countries will have been covered by seamless 30 cm GSD color-, 60 cm GSD false-color infrared ortho-mosaics and a 1 meter GSD digital surface model. The ortho-maps are being published to Microsoft's Bing Maps Internet mapping portal. The Global Ortho Program was designed for highly and unprecedented automated mapping of essentially entire continents. In 2011, exclusive of flight operations, the product output per person has been measured in excess of 275,000 square km per year. We describe research efforts that made this achievement possible. Those include a specially designed aerial sensor (Ultracam G), logistics simulation for fight planning and optimization, in-flight blur detection and subsequent automatic blur removal, modeling and removal of atmospheric and environmental conditions, automated shear detection and DTM refinement, an IT architecture to process >200,000 aerial images/day, and for creating over 1,000,000 km2 ortho-imagery and DSM data in 24 hours. While addressing these issues, we provide ideas how this might affect the future of spatial infrastructure initiatives.

Potential effects in multi-resolution post-classification change detection

Change detection is one of the primary applications of remote-sensing data, and many techniques have been developed during the past three decades. Although frequently criticized and despite many alternatives, due to its simplicity and intuitive manner, post-classification change detection still remains one of the most applied techniques. Many studies in the field of change detection analysis acknowledge, for instance, the impact of misregistration, inconsistencies in classification schemes or differences in methods for image interpretation. However, there are additional, rarely studied influences that can cause large errors in change detection results, including integrating multi-resolution data, the adjacency effect and the minimum mapping units (MMUs) that are applied to the classification before change detection. This study demonstrates these effects for the complex land cover of the Alvarado mangrove area at the Mexican Gulf Coast, employing 10 m Système Pour l'Observation de la Terre 5 (SPOT-5) high geometric resolution (HRG)‐based and 57 m Landsat Multispectral Scanner (MSS) classifications. As a baseline, the proportion of the fine spatial resolution classes within the coarse spatial resolution cells were derived, from which proportional change matrices were computed. The analysis employs difference measures to compare change matrices and proportional maps. The impact of various tested resampling functions was negligible if coarse resolution data were refined. For coarsening fine spatial resolution data, change matrix comparison was comparatively small but yielded differences of approximately 20% in spatially explicit measures. Incorrect positional alignment indicated differences of up to 5% in the change matrix for a misregistration of 100 m and even higher spatially explicit differences (28%). The discrepancies due to the adjacency effect were rather low. MMUs of 25 ha resulted in differences of up to 36% in the change matrix. The magnitude of the discrepancies of all studied effects depends on the class diversity in the map, and some can also be related to the difference in spatial resolution.

Tree species classification in the Southern Alps based on the fusion of very high geometrical resolution multispectral/hyperspectral images and LiDAR data

The identification of tree species is an important issue in forest management. In recent years, many studies have explored this topic using hyperspectral, multispectral, and LiDAR data. In this study we analyzed two multi-sensor set-ups: 1) airborne high spatial resolution hyperspectral images combined with LiDAR data; and 2) high spatial resolution satellite multispectral images combined with LiDAR data. Two LiDAR acquisitions were considered: low point density (approx. 0.48points per m2) and high point density (approx. 8.6points per m2). The aims of this work were: i) to understand what level of classification accuracy can be achieved using a high spectral and spatial resolution multi-sensor data set-up (very high spatial and spectral resolution airborne hyperspectral images integrated with high point density LiDAR data), over a mountain area characterized by many species, both broadleaf and coniferous; ii) to understand the implications of a downgrading of the data characteristics (in terms of spectral resolution of spectral data and point density of LiDAR data), on species separability, with respect to the previous set-up; and iii) to understand the differences between high- and low-point density LiDAR acquisitions on tree species classification. The study region was a mountain area in the Southern Alps characterized by many tree species (7 species and a “non-forest” class), either coniferous or broadleaf. For each set-up a specific processing chain was adopted, from the pre-processing of the raw data to the classification (two classifiers were used: support vector machine and random forest). Different class definitions were tested, including general macro-classes, forest types, and finally single tree species. Experimental results showed that the set-up based on hyperspectral data was effective with general macro-classes, forest types, and single species, reaching high kappa accuracies (93.2%, 82.1% and 76.5%, respectively). The use of multispectral data produced a reduction in the classification accuracy, which was sharp for single tree species, and still high for forest types. Considering general macro-classes, the multispectral set-up was still very accurate (85.8%). Regarding LiDAR data, the experimental analysis showed that high density LiDAR data provided more information for tree species classification with respect to low density data, when combined with either hyperspectral or multispectral data.

Supervised change detection in VHR images using contextual information and support vector machines

In this paper we study an effective solution to deal with supervised change detection in very high geometrical resolution (VHR) images. High within-class variance as well as low between-class variance that characterize this kind of imagery make the detection and classification of ground cover transitions a difficult task. In order to achieve high detection accuracy, we propose the inclusion of spatial and contextual information issued from local textural statistics and mathematical morphology. To perform change detection, two architectures, initially developed for medium resolution images, are adapted for VHR: Direct Multi-date Classification and Difference Image Analysis. To cope with the high intra-class variability, we adopted a nonlinear classifier: the Support Vector Machines (SVM). The proposed approaches are successfully evaluated on two series of pansharpened QuickBird images.

Multichannel Azimuth Processing in ScanSAR and TOPS Mode Operation

Due to a system-inherent limitation, conventional synthetic aperture radar (SAR) is incapable of imaging a wide swath with high geometric resolution. This restriction can be overcome by systems with multiple receive channels in combination with an additional digital signal processing network. So far, the application of such digital beamforming algorithms for high-resolution wide-swath SAR imaging has been restricted to multichannel systems in stripmap operation. However, in stripmap mode, the overall azimuth antenna length restricts the achievable swath width, thus preventing very wide swaths as requested by future SAR missions. Consequently, new concepts for ultrawide-swath imaging are needed. A promising candidate is a SAR system with multiple azimuth channels being operated in burst mode. This paper analyzes innovative ScanSAR and Terrain Observation by Progressive Scans (TOPS) system concepts with regard to multichannel azimuth processing. For this, the theoretical analyses, performance figures, and SAR signal processing, which had previously been derived for multichannel stripmap mode, are extended to systems operating in burst modes. The investigations reveal that multichannel ScanSAR systems enable the imaging of ultrawide swaths with high azimuth resolution and compact antenna lengths. These considerations are embedded in a multichannel ScanSAR system design example to demonstrate its capability to image an ultrawide swath of 400 km with a high geometric resolution of 5 m. In a next step, this system is adapted to TOPS mode operation, including an innovative “staircase” multichannel processing approach optimized for TOPS.

A Context-Sensitive Technique Robust to Registration Noise for Change Detection in VHR Multispectral Images

This paper presents an automatic context-sensitive technique robust to registration noise (RN) for change detection (CD) in multitemporal very high geometrical resolution (VHR) remote sensing images. Exploiting the properties of RN in VHR images, the proposed technique analyzes the distribution of the spectral change vectors (SCVs) computed according to the change vector analysis (CVA) in a quantized polar domain. The method studies the SCVs falling into each quantization cell at different resolution levels (scales) to automatically identify the effects of RN in the polar domain. This information is jointly exploited with the spatial context information contained in the neighborhood of each pixel for generating the final CD map. The spatial context information is modeled through the definition of adaptive regions homogeneous both in spatial and temporal domain (parcels). Experimental results obtained on real VHR remote sensing multitemporal images confirm the effectiveness of the proposed technique.

Metrical use of declassified satellite imagery for an area of archaeological interest in Turkey

Abstract The adoption of satellite imagery in archaeology is well recognized today, and is rapidly increasing with the greater availability of high resolution sensors, in the past limited solely to military applications. The metrical use of declassified satellite images from the intelligence CORONA project, the first US operational space photo reconnaissance project operating in the period 1960–1972, has been verified within the framework of a multidisciplinary research project carried out by the University of Bologna, in collaboration with Istanbul University and Gaziantep archaeological museum, at the archaeological site of Tilmen Hoyuk (south-eastern Turkey) and its surrounding landscape. These images, characterized by a panoramic geometry, indeed represent an important historical resource for photointerpretation in archaeology, but their metrical use is difficult, owing to the severe distortions affecting the acquisition together with the problems of finding technical data related to the missions. In order to produce metric products of sufficient quality for mapping purposes, a high resolution scanned copy of a CORONA image of the relevant area was processed using commercial digital photogrammetric workstations operating with different approaches, whose aim was to evaluate the potential and specific problems connected to these kinds of data. A full orientation and digital triangulation of these images is possible if all the technical data on the sensor and the acquisition are well-known, and in any case the existence of well-distributed and reliable ground control points in the study area is crucial. The results obtained are interesting as they allow for integration, on a common cartographic reference base, with current aerial and satellite imagery (medium to very high geometric resolution) to perform automatic or visual change analysis procedures.

Optimisation of LiDAR derived terrain models for river flow modelling

Abstract. Airborne LiDAR (Light Detection And Ranging) combines cost efficiency, high degree of automation, high point density of typically 1–10 points per m2 and height accuracy of better than ±15 cm. For all these reasons LiDAR is particularly suitable for deriving precise Digital Terrain Models (DTM) as geometric basis for hydrodynamic-numerical (HN) simulations. The application of LiDAR for river flow modelling requires a series of preprocessing steps. Terrain points have to be filtered and merged with river bed data, e.g. from echo sounding. Then, a smooth Digital Terrain Model of the Watercourse (DTM-W) needs to be derived, preferably considering the random measurement error during surface interpolation. In a subsequent step, a hydraulic computation mesh has to be constructed. Hydraulic simulation software is often restricted to a limited number of nodes and elements, thus, data reduction and data conditioning of the high resolution LiDAR DTM-W becomes necessary. We will present a DTM thinning approach based on adaptive TIN refinement which allows a very effective compression of the point data (more than 95% in flood plains and up to 90% in steep areas) while preserving the most relevant topographic features (height tolerance ±20 cm). Traditional hydraulic mesh generators focus primarily on physical aspects of the computation grid like aspect ratio, expansion ratio and angle criterion. They often neglect the detailed shape of the topography as provided by LiDAR data. In contrast, our approach considers both the high geometric resolution of the LiDAR data and additional mesh quality parameters. It will be shown that the modelling results (flood extents, flow velocities, etc.) can vary remarkably by the availability of surface details. Thus, the inclusion of such geometric details in the hydraulic computation meshes is gaining importance in river flow modelling.