on-Board surround sensors, such as cameras and radars, are used in Advanced Driving Assistance Systems applications to improve the driver safety and comfort. In the meantime, Car-to-Car Communication systems are in the deployment phase and have been tested in huge test fields. In order to compensate the weaknesses and benefit from the strengths of both systems, their information can be fused. In this context, one major challenge is the unambiguous assignment of detected vehicles from different sensors, which is still an open research topic and the major target of this work. An innovative algorithm was first tested in Matlab using recorded data and then implemented on real hardware. The results obtained are promising and, from a spatial point of view, they show already a successful matching of vehicles. Compared with the solutions proposed in literature, the developed demonstrator is innovative, and represents the first step towards a real world application running in real time inside cars. Overall, this work is a useful contribution to active safety and autonomous driving applications based on sensor fusion and a good reference for further research on the topic.

With the exponential increasing demands and uses of GIS data visualization system, such as urban planning, environment and climate change monitoring, weather simulation, hydrographic gauge and so forth, the geospatial vector and raster data visualization research, application and technology has become prevalent. However, we observe that current web GIS techniques are merely suitable for static vector and raster data where no dynamic overlaying layers. While it is desirable to enable visual explorations of large-scale dynamic vector and raster geospatial data in a web environment, improving the performance between backend datasets and the vector and raster applications remains a challenging technical issue. This dissertation is to implement these challenging and unimplemented areas: how to provide a large-scale dynamic vector and raster data visualization service with dynamic overlaying layers accessible from various client devices through a standard web browser, and how to make the large-scale dynamic vector and raster data visualization service as rapid as the static one. To accomplish these, a large-scale dynamic vector and raster data visualization geographic information system based on parallel map tiling and a comprehensive performance improvement solution are proposed, designed and implemented. They include: the quadtree-based indexing and parallel map tiling, the Legend String, the vector data visualization with dynamic layers overlaying, the vector data time series visualization, the algorithm of vector data rendering, the algorithm of raster data re-projection, the algorithm for elimination of superfluous level of detail, the algorithm for vector data gridding and re-grouping and the cluster servers side vector and raster data caching.

With the emergence of innovative applications for Virtual Reality (VR) in touring, E-commerce, and social activities, high-quality VR video streaming becomes essential. To support numerous wireless VR users, this paper aims to leverage video synthesis techniques to effectively reduce the multicast bandwidth consumption. It synthesizes the view in the video for a user from the one of a nearby user with similar Field of View (FoV), under the virtual distance and view angle constraints. We first formulate a new optimization problem, named VR Content Sharing and Multicasting (VCSM), and prove the NP-hardness. Then, we propose View Sharing Relation Graph (VSRG) to model the synthesis relation between each pair of views. We then design a new algorithm, named Bandwidth-Efficient Multicast with Synthesis (BEMS) to select multicast views and the corresponding MCS in wireless networks. Simulation results show that BEMS can effectively reduce bandwidth consumption by more than 50% compared with state-of-the-art wireless transmission schemes.

Wireless virtual reality (VR) applications that provide users extraordinary viewing experience are now drawing great attentions. Transmitting VR video via wireless channel to users’ head-mounted display devices efficiently with low latency is very important for many emerging VR applications. In this paper, we propose a pseudo-analog transmission framework named OmniCast, which provides graceful quality degradation and competitive performance for unpredictably varying wireless channels while featuring low latency, low complexity, and low energy cost. In particular, we analyze the influence of projection between the spherical representation and the 2-D plane representation, and derive a power optimization scheme to minimize the distortion on the sphere. In addition, we develop an approach to measure the efficiency of decorrelation transform in the spherical domain. Based on that, an appropriate transform option can be determined. Experimental results show that the proposed framework improves the transmission efficiency of omnidirectional videos, while achieving elegant quality degradation for channel fluctuation in a wide channel SNR range.

Wide-angle images gained a huge popularity in the last years due to the development of computational photography and imaging technological advances. They present the information of a scene in a way which is more natural for the human eye but, on the other hand, they introduce artifacts such as bent lines. These artifacts become more and more unnatural as the field of view increases. In this work, we present a technique aimed to improve the perceptual quality of panorama visualization. The main ingredients of our approach are, on one hand, considering the viewing sphere as a Riemann sphere, what makes natural the application of Möbius (complex) transformations to the input image, and, on the other hand, a projection scheme which changes in function of the field of view used. We also introduce an implementation of our method, compare it against images produced with other methods and show that the transformations can be done in real time, which makes our technique very appealing for new settings, as well as for existing interactive panorama applications.

While the human hearing capability for horizontally localized sound sources is mostly based on binaural cues, the localization of elevation along the "cones of confusion" can only rely on spectral cues provided by the directionality of head related transfer functions (HRTF). This paper explores how the magnitude of spectral differences between HRTF affects the availability of spectral cues to perceive and detect different elevations and hence limits localization accuracy. A method to predict localization accuracy is presented, based on analysis of the gradient of HRTF log-spectral differences for elevation positional changes. To this end, the localization accuracy is estimated from an HRTF database with 45 subjects. Validation of the results shows good agreement with results from subjective localization experiments reported in literature.

Weather forecasting is a problem where an enormous amount of data must be processed. Severe storms cause a significant amount of damages and loss every year in part due to the insufficiency of the current techniques in producing reliable forecasts. We propose an algorithm that analyzes satellite images from the vast historical archives to predict severe storms. Conventional weather forecasting involves solving numerical models based on sensory data. It has been challenging for computers to make forecasts based on the visual patterns from satellite images. In our system we extract and summarize important visual storm evidence from satellite image sequences in a way similar to how meteorologists interpret these images. Particularly, the algorithm extracts and fits local cloud motions from image sequences to model the storm-related cloud patches. Image data of an entire year are adopted to train the model. The historical storm reports since the year 2000 are used as the ground-truth and statistical priors in the modeling process. Experiments demonstrate the usefulness and potential of the algorithm for producing improved storm forecasts.

We use South Pole Telescope data from 2008 and 2009 to detect the non-Gaussian signature in the cosmic microwave background (CMB) produced by gravitational lensing and to measure the power spectrum of the projected gravitational potential. We constrain the ratio of the measured amplitude of the lensing signal to that expected in a fiducial ΛCDM cosmological model to be 0.86 ± 0.16, with no lensing disfavored at 6.3σ. Marginalizing over ΛCDM cosmological models allowed by the Wilkinson Microwave Anisotropy Probe (WMAP7) results in a measurement of Alens = 0.90 ± 0.19, indicating that the amplitude of matter fluctuations over the redshift range 0.5 ≲ z ≲ 5 probed by CMB lensing is in good agreement with predictions. We present the results of several consistency checks. These include a clear detection of the lensing signature in CMB maps filtered to have no overlap in Fourier space, as well as a “curl” diagnostic that is consistent with the signal expected for ΛCDM. We perform a detailed study of bias in the measurement due to noise, foregrounds, and other effects and determine that these contributions are relatively small compared to the statistical uncertainty in the measurement. We combine this lensing measurement with results from WMAP7 to improve constraints on cosmological parameters when compared to those from WMAP7 alone: we find a factor of 3.9 improvement in the measurement of the spatial curvature of the universe, Ωk = −0.0014 ± 0.0172; a 10% improvement in the amplitude of matter fluctuations within ΛCDM, σ8 = 0.810 ± 0.026; and a 5% improvement in the dark energy equation of state, w = −1.04 ± 0.40. When compared with the measurement of w provided by the combination of WMAP7 and external constraints on the Hubble parameter, the addition of the lensing data improves the measurement of w by 15% to give w = −1.087 ± 0.096.

We propose a novel framework for accurate 3D georegistration of wide area motion imagery (WAMI), which is a challenging problem because parametric transformations are insufficient for aligning WAMI image frames to a georeferenced coordinate system in urban areas containing tall buildings and 3D structures. Using structure from motion (SfM) we estimate a 3D point cloud for the scene. Independently, we also compute a precise alignment between the roads in the WAMI frames and a georeferenced vector roadmap by detecting locations of moving vehicles and aligning these locations with the roads in the vector roadmap via parametric chamfer matching. The aligned vector roadmap then identifies corresponding pixels in the WAMI frames, which can be triangulated using the SfM camera parameters to obtain a set of sparse but georeferenced points in the SfM 3D coordinate frame that directly enable georegistration of the complete 3D scene point cloud via a similarity transform. The proposed methodology enables 3D georegistration of a sequence of WAMI frames using only georeferenced vector roadmaps, which are readily available, and without requiring independent georeferenced lidar scans that have been used in prior work. Our framework is validated on WAMI dataset including high resolution WAMI frames for the downtown Rochester, NY region. Experimental results demonstrate that the proposed framework produces an accurate georeferenced point cloud representation for the scene.

We propose a kernel-based framework for computing components from a set of surface normals. This framework allows us to easily demonstrate that component analysis can be performed directly upon normals. We link previously proposed mapping functions, the azimuthal equidistant projection (AEP) and principal geodesic analysis (PGA), to our kernel-based framework. We also propose a new mapping function based upon the cosine distance between normals. We demonstrate the robustness of our proposed kernel when trained with noisy training sets. We also compare our kernels within an existing shape-from-shading (SFS) algorithm. Our spherical representation of normals, when combined with the robust properties of cosine kernel, produces a very robust subspace analysis technique. In particular, our results within SFS show a substantial qualitative and quantitative improvement over existing techniques.

We present imaging and high resolution energy spectra of energetic ∼5–90 keV neutral atoms (ENA) of a weak geomagnetic substorm (Dst > −8 nT and AE ≲ 200 nT), made by the Suprathermal Electron (STE) instrument on the STEREO B spacecraft. Enhanced ENA emissions were observed coming from around local midnight near the equator with different spatial distribution and/or temporal behavior at ∼5–20keV compared to ∼20–90 keV. By forward modeling using a parameterized ring‐current model, we show that the ENA images imply the parent equatorial protons have pitch‐angle distributions peaked at 90°, an energy spectrum consistent with in situ proton measurements at geosynchronous orbit, and a spatial asymmetry with the maximum flux at midnight for 5–20 keV and at 2240 MLT for 20–90 keV. These are the first ENA measurements at ∼5 to 26 keV from low altitude, and the first for such weak activity.

We present and evaluate the capacity of a deep neural network to learn robust features from EEG to automatically detect seizures. This is a challenging problem because seizure manifestations on EEG are extremely variable both inter- and intra-patient. By simultaneously capturing spectral, temporal and spatial information our recurrent convolutional neural network learns a general spatially invariant representation of a seizure. The proposed approach exceeds significantly previous results obtained on cross-patient classifiers both in terms of sensitivity and false positive rate. Furthermore, our model proves to be robust to missing channel and variable electrode montage.

We present an approach to systematically analysing the vulnerability of road networks under disruptions covering extended areas. Since various kinds of events including floods, heavy snowfall, storms and wildfires can cause such spatially spread degradations, the analysis method is an important complement to the existing studies of single link failures. The methodology involves covering the study area with grids of uniformly shaped and sized cells, where each cell represents the extent of an event disrupting any intersecting links. We apply the approach to the Swedish road network using travel demand and network data from the Swedish national transport modelling system Sampers. The study shows that the impacts of area-covering disruptions are largely determined by the level of internal, outbound and inbound travel demand of the affected area itself. This is unlike single link failures, where the link flow and the redundancy in the surrounding network determine the impacts. As a result, the vulnerability to spatially spread events shows a markedly different geographical distribution. These findings, which should be universal for most road networks of similar scale, are important in the planning process of resource allocation for mitigation and recovery.

We present an algorithm for converting an indoor spherical panorama into a photograph with a simulated overhead view. The resulting image will have an extremely wide field of view covering up to 4{\pi} steradians of the spherical panorama. We argue that our method complements the stereographic projection commonly used in the "little planet" effect. The stereographic projection works well in creating little planets of outdoor scenes; whereas our method is a well-suited counterpart for indoor scenes. The main innovation of our method is the introduction of a novel azimuthal map projection that can smoothly blend between the stereographic projection and the Lambert azimuthal equal-area projection. Our projection has an adjustable parameter that allows one to control and compromise between distortions in shape and distortions in size within the projected panorama. This extra control parameter gives our projection the ability to produce superior results over the stereographic projection.

We present a novel technique for the automatic alignment of Structure from Motion (SfM) models, acquired at ground level or by micro aerial vehicles, to an overhead Digital Surface Model (DSM) using GPS information. An additional refinement step based on the correlation of the DSM height map with the model height map corrects for the GPS localization uncertainties and results in precisely aligned models. Our approach successfully handles cases where previous methods had problems, including objects on the ground, unoccupied space, and models covering a small area. We conclude our work by presenting several applications of our approach, namely the fusion of detailed SfM model information into the original DSM, season-invariant matching using aligned models, and alignment for providing context in visualization.

We present a new projection format and packing layout for 360 VR videos using octahedron mapping. A spherical video is projected onto an octahedron, where the upper and lower hemispheres correspond to the upper and lower half of the octahedron, respectively. Four regular triangles in the half of the octahedron, are transformed into isosceles right-angled triangles and packed into a square by remaining adjacent edges. Two equal squares resulted respectively from the upper and lower half of the octahedron, are placed side by side and adjoined by a common edge. This generates a 2:1 aspect ratio rectangle. We demonstrate that our new projection format and layout have advantages in uniformity of pixel density, internal continuity and rectangle aspect ratio while encoding 360 VR videos.

We present a new approach that automatically estimates global pose for a UAV in real-time using 3D terrain engine. Inaccurate auxiliary sensors on the UAV were used to obtain initial real camera pose that moves the virtual camera inside the 3D terrain engine. We, then automatically found multiple matches between the two images to find the 3D coordinates of the matches using the 3D terrain engine. Finally, we tested the co-planarity of the 3D points under the camera, depending on this test. We used coplanar or non-coplanar algorithm to estimate accurate global camera pose. We executed feature detection, description and pair wise matching algorithms on GPU to get a suitable frame rate (12 FPS) needed in the navigation applications. The proposed approach has been tested on a synthetic and real data. Experimental results proved the feasibility and robustness of the proposed approach, and the precision was the same order as the 3D terrain engine used. Finally, we can say that the 3D terrain engine succeeded when other methods failed.

We present a method for manipulation and visualization of wide-angle images using transformations defined on the complex plane C. We map the unit sphere S2 to C using the stereographic projection, multiply the complex plane by a given complex number, and map the result back to the sphere using the inverse of the stereographic projection. Since all these transformations preserve angle, we obtain a result containing only distortions due to the latitude/longitude representation of the sphere, which were already present in the input image. We then explore the possibility given by our technique of mapping wide fields of view to narrower ones. This makes possible to apply perspective projection to wider fields of view, leading to a natural generalization of the perspective projection in the context of panoramic images. Our results are generated in real-time and compare competitively with state-of-the-art methods used to project the viewing sphere to the image plane.

We present a measurement of the cosmic microwave background (CMB) temperature power spectrum using data from the recently completed South Pole Telescope Sunyaev-Zel'dovich (SPT-SZ) survey. This measurement is made from observations of 2540 deg^2 of sky with arcminute resolution at 150 GHz, and improves upon previous measurements using the SPT by tripling the sky area. We report CMB temperature anisotropy power over the multipole range 650 < l < 3000. We fit the SPT bandpowers, combined with the 7 yr Wilkinson Microwave Anisotropy Probe (WMAP7) data, with a six-parameter ΛCDM cosmological model and find that the two datasets are consistent and well fit by the model. Adding SPT measurements significantly improves ΛCDM parameter constraints; in particular, the constraint on θ_s tightens by a factor of 2.7. The impact of gravitational lensing is detected at 8.1σ, the most significant detection to date. This sensitivity of the SPT+WMAP7 data to lensing by large-scale structure at low redshifts allows us to constrain the mean curvature of the observable universe with CMB data alone to be Ω_k=-0.003^(+0.014)_(-0.018). Using the SPT+WMAP7 data, we measure the spectral index of scalar fluctuations to be n_s = 0.9623 ± 0.0097 in the ΛCDM model, a 3.9σ preference for a scale-dependent spectrum with n_s < 1. The SPT measurement of the CMB damping tail helps break the degeneracy that exists between the tensor-to-scalar ratio r and n_s in large-scale CMB measurements, leading to an upper limit of r < 0.18 (95% C.L.) in the ΛCDM+r model. Adding low-redshift measurements of the Hubble constant (H_0) and the baryon acoustic oscillation (BAO) feature to the SPT+WMAP7 data leads to further improvements. The combination of SPT+WMAP7+H_0+BAO constrains n_s = 0.9538 ± 0.0081 in the ΛCDM model, a 5.7σ detection of n_s < 1, and places an upper limit of r < 0.11 (95% C.L.) in the ΛCDM+r model. These new constraints on n_s and r have significant implications for our understanding of inflation, which we discuss in the context of selected single-field inflation models.

We present a measurement of the angular power spectrum of the cosmic microwave background (CMB) using data from the South Pole Telescope (SPT). The data consist of 790 square degrees of sky observed at 150 GHz during 2008 and 2009. Here we present the power spectrum over the multipole range 650 < ‘ < 3000, where it is dominated by primary CMB anisotropy. We combine this power spectrum with the power spectra from the seven-year Wilkinson Microwave Anisotropy Probe (WMAP) data release to constrain cosmological models. We nd that the SPT and WMAP data are consistent with each other and, when combined, are well t by a spatially at, CDM cosmological model. The SPT+WMAP constraint on the spectral index of scalar uctuations is ns = 0:9663 0:0112. We detect, at 5 signicance, the eect of gravitational lensing on the CMB power spectrum, and nd its amplitude to be consistent with the CDM cosmological model. We explore a number of extensions beyond the CDM model. Each extension is tested independently, although there are degeneracies between some of the extension parameters. We constrain the tensorto-scalar ratio to be r < 0:21 (95% CL) and constrain the running of the scalar spectral index to be dns=d lnk = 0:024 0:013. We strongly detect the eects of primordial helium and neutrinos on the CMB; a model without helium is rejected at 7.7 , while a model without neutrinos is rejected at 7.5 . The primordial helium abundance is measured to be Yp = 0:296 0:030, and the eective number of relativistic species is measured to be Ne = 3:85 0:62. The constraints on these models are strengthened when the CMB data are combined with measurements of the Hubble constant and the baryon acoustic oscillation feature. Notable improvements include ns = 0:9668 0:0093, r < 0:17 (95% CL), and Ne = 3:86 0:42. The SPT+WMAP data show a mild preference for low power in the CMB damping tail, and while this preference may be accommodated by models that have a negative spectral running, a high primordial helium abundance, or a high eective number of relativistic species, such models are disfavored by the abundance of low-redshift galaxy clusters. Subject headings: cosmology { cosmology:cosmic microwave background { cosmology: observations { large-scale structure of universe