ALBERT

Beschreibung: The use of hyperspectral data to determine the abundance of constituents in a certain portion of the Earth's surface relies on the capability of imaging spectrometers to provide a large amount of information at each pixel of a certain scene. Today, hyperspectral imaging sensors are capable of generating unprecedented volumes of radiometric data. The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), for example, routinely produces image cubes with 224 spectral bands. This undoubtedly opens a wide range of new possibilities, but the analysis of such a massive amount of information is not an easy task. In fact, most of the existing algorithms devoted to analyzing multispectral images are not applicable in the hyperspectral domain, because of the size and high dimensionality of the images. The application of neural networks to perform unsupervised classification of hyperspectral data has been tested by several authors and also by us in some previous work. We have also focused on analyzing the intrinsic capability of neural networks to parallelize the whole hyperspectral unmixing process. The results shown in this work indicate that neural network models are able to find clusters of closely related hyperspectral signatures, and thus can be used as a powerful tool to achieve the desired classification. The present work discusses the possibility of using a Self Organizing neural network to perform unsupervised classification of hyperspectral images. In sections 3 and 4, the topology of the proposed neural network and the training algorithm are respectively described. Section 5 provides the results we have obtained after applying the proposed methodology to real hyperspectral data, described in section 2. Different parameters in the learning stage have been modified in order to obtain a detailed description of their influence on the final results. Finally, in section 6 we provide the conclusions at which we have arrived.

Beschreibung: During the last several years, a number of airborne and satellite hyperspectral sensors have been developed or improved for remote sensing applications. Imaging spectrometry allows the detection of materials, objects and regions in a particular scene with a high degree of accuracy. Hyperspectral data typically consist of hundreds of thousands of spectra, so the analysis of this information is a key issue. Mathematical morphology theory is a widely used nonlinear technique for image analysis and pattern recognition. Although it is especially well suited to segment binary or grayscale images with irregular and complex shapes, its application in the classification/segmentation of multispectral or hyperspectral images has been quite rare. In this paper, we discuss a new completely automated methodology to find endmembers in the hyperspectral data cube using mathematical morphology. The extension of classic morphology to the hyperspectral domain allows us to integrate spectral and spatial information in the analysis process. In Section 3, some basic concepts about mathematical morphology and the technical details of our algorithm are provided. In Section 4, the accuracy of the proposed method is tested by its application to real hyperspectral data obtained from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) imaging spectrometer. Some details about these data and reference results, obtained by well-known endmember extraction techniques, are provided in Section 2. Finally, in Section 5 we expose the main conclusions at which we have arrived.

Beschreibung: The GeoEye Constellation consists of: a) IKONOS and OrbView-3 for high resolution; b) GeoEye with higher resolution 1Q2007; c) RESOUCESAT-1 for global crop assessment; d) OrbView-2 for ocean research and fish. IKONOS performance in 2005 included stable image quality, radiometry and geometric accuracy. reliability is 80% to 2008. Demonstrated capacity for high-volume, quick-response collection and production.

Beschreibung: Airborne laser swath mapping (ALSM) of the Puget Lowland conducted by TerraPoint LLC for the Purget Sound Lidar Concortium (PSLC), has been successful in revealing Holocene fault scarps and lendsliders hidden beneath the dense, temperate rain forest cover and in quantifying shoreline terrace uplift. Expanding the PSLC efforts, NASA-USGS collaboration is now focusing on topographic mapping of seismogenic zones adjacent to volcanois in the western Cascades range in order to assess the presence of active faulting and tectonic deformation, better define the extend of lahars and understand their flow processes, and characterize landslide occurrence. Mapping of the western Rainier zone (WRZ) was conducted by TerraPoint in late 2002, after leaf fall and before snow accumulation. The WRZ is a NNW-trending, approx. 30 km-long zone of seismicity west of Mount Rainier National Park. The Puget Lowland ALSM methods were modified to accommodate challenges posed by the steep, high relief terrian. The laser data, acquired with a density of approx. 2 pulses /sq m, was filtered to identify returns from the ground from which a bare Earth digital elevation model (DEM) was produced with a grid size of 1.8 m. The RMS elevation accuracy of the DEM in flat, unvegetated areas is approx. 10cm based on consistency between overlapping flight swaths and comparisons to ground control points. The resulting DEM substantially improves upon Shuttle Radar Topography Mission and USGS photogrammetric mapping. For example, the DEM defines the size and spatial distribution of flood erratics left by the Electron lahar and of megaclasts within the Round Pass lahar, important for characterizing the lahar hydraulics. A previously unknown lateral levee on the Round Pass lahar is also revealed. In addition, to illustrating geomorfic feature within the WRZ, future plans for laser mapping of the Saint Helens and Darrington seismic zones will be described.

Beschreibung: This viewgraph presentation reviews the use of Image Tour to explore the multiangle, multispectral satellite imagery. Remote sensing data are spatial arrays of p-dimensional vectors where each component corresponds to one of p variables. Applying the same R(exp p) to R(exp d) projection to all pixels creates new images, which may be easier to analyze than the original because d 〈 p. Image grand tour (IGT) steps through the space of projections, and d=3 outputs a sequence of RGB images, one for each step. In this talk, we apply IGT to multiangle, multispectral data from NASA's MISR instrument. MISR views each pixel in four spectral bands at nine view angles. Multiple views detect photon scattering in different directions and are indicative of physical properties of the scene. IGT allows us to explore MISR's data structure while maintaining spatial context; a key requirement for physical interpretation. We report results highlighting the uniqueness of multiangle data and how IGT can exploit it.