The Experts below are selected from a list of 15330 Experts worldwide ranked by ideXlab platform
Davide Scaramuzza - One of the best experts on this subject based on the ideXlab platform.
-
EKLT: Asynchronous Photometric Feature Tracking Using Events and Frames
International Journal of Computer Vision, 2020Co-Authors: Daniel Gehrig, Henri Rebecq, Guillermo Gallego, Davide ScaramuzzaAbstract:We present EKLT, a feature tracking method that leverages the complementarity of event cameras and standard cameras to track visual features with high temporal resolution. Event cameras are novel sensors that Output Pixel-level brightness changes, called “events”. They offer significant advantages over standard cameras, namely a very high dynamic range, no motion blur, and a latency in the order of microseconds. However, because the same scene pattern can produce different events depending on the motion direction, establishing event correspondences across time is challenging. By contrast, standard cameras provide intensity measurements (frames) that do not depend on motion direction. Our method extracts features on frames and subsequently tracks them asynchronously using events, thereby exploiting the best of both types of data: the frames provide a photometric representation that does not depend on motion direction and the events provide updates with high temporal resolution. In contrast to previous works, which are based on heuristics, this is the first principled method that uses intensity measurements directly, based on a generative event model within a maximum-likelihood framework. As a result, our method produces feature tracks that are more accurate than the state of the art, across a wide variety of scenes.
-
ultimate slam combining events images and imu for robust visual slam in hdr and high speed scenarios
International Conference on Robotics and Automation, 2018Co-Authors: Antonio Rosinol Vidal, Henri Rebecq, Timo Horstschaefer, Davide ScaramuzzaAbstract:Event cameras are bioinspired vision sensors that Output Pixel-level brightness changes instead of standard intensity frames. These cameras do not suffer from motion blur and have a very high dynamic range, which enables them to provide reliable visual information during high-speed motions or in scenes characterized by high dynamic range. However, event cameras Output only little information when the amount of motion is limited, such as in the case of almost still motion. Conversely, standard cameras provide instant and rich information about the environment most of the time (in low-speed and good lighting scenarios), but they fail severely in case of fast motions, or difficult lighting such as high dynamic range or low light scenes. In this letter, we present the first state estimation pipeline that leverages the complementary advantages of these two sensors by fusing in a tightly coupled manner events, standard frames, and inertial measurements. We show on the publicly available Event Camera Dataset that our hybrid pipeline leads to an accuracy improvement of 130% over event-only pipelines, and 85% over standard-frames-only visual-inertial systems, while still being computationally tractable. Furthermore, we use our pipeline to demonstrate-to the best of our knowledge-the first autonomous quadrotor flight using an event camera for state estimation, unlocking flight scenarios that were not reachable with traditional visual-inertial odometry, such as low-light environments and high dynamic range scenes. Videos of the experiments: http://rpg.ifi.uzh.ch/ultimateslam.html.
Henri Rebecq - One of the best experts on this subject based on the ideXlab platform.
-
EKLT: Asynchronous Photometric Feature Tracking Using Events and Frames
International Journal of Computer Vision, 2020Co-Authors: Daniel Gehrig, Henri Rebecq, Guillermo Gallego, Davide ScaramuzzaAbstract:We present EKLT, a feature tracking method that leverages the complementarity of event cameras and standard cameras to track visual features with high temporal resolution. Event cameras are novel sensors that Output Pixel-level brightness changes, called “events”. They offer significant advantages over standard cameras, namely a very high dynamic range, no motion blur, and a latency in the order of microseconds. However, because the same scene pattern can produce different events depending on the motion direction, establishing event correspondences across time is challenging. By contrast, standard cameras provide intensity measurements (frames) that do not depend on motion direction. Our method extracts features on frames and subsequently tracks them asynchronously using events, thereby exploiting the best of both types of data: the frames provide a photometric representation that does not depend on motion direction and the events provide updates with high temporal resolution. In contrast to previous works, which are based on heuristics, this is the first principled method that uses intensity measurements directly, based on a generative event model within a maximum-likelihood framework. As a result, our method produces feature tracks that are more accurate than the state of the art, across a wide variety of scenes.
-
ultimate slam combining events images and imu for robust visual slam in hdr and high speed scenarios
International Conference on Robotics and Automation, 2018Co-Authors: Antonio Rosinol Vidal, Henri Rebecq, Timo Horstschaefer, Davide ScaramuzzaAbstract:Event cameras are bioinspired vision sensors that Output Pixel-level brightness changes instead of standard intensity frames. These cameras do not suffer from motion blur and have a very high dynamic range, which enables them to provide reliable visual information during high-speed motions or in scenes characterized by high dynamic range. However, event cameras Output only little information when the amount of motion is limited, such as in the case of almost still motion. Conversely, standard cameras provide instant and rich information about the environment most of the time (in low-speed and good lighting scenarios), but they fail severely in case of fast motions, or difficult lighting such as high dynamic range or low light scenes. In this letter, we present the first state estimation pipeline that leverages the complementary advantages of these two sensors by fusing in a tightly coupled manner events, standard frames, and inertial measurements. We show on the publicly available Event Camera Dataset that our hybrid pipeline leads to an accuracy improvement of 130% over event-only pipelines, and 85% over standard-frames-only visual-inertial systems, while still being computationally tractable. Furthermore, we use our pipeline to demonstrate-to the best of our knowledge-the first autonomous quadrotor flight using an event camera for state estimation, unlocking flight scenarios that were not reachable with traditional visual-inertial odometry, such as low-light environments and high dynamic range scenes. Videos of the experiments: http://rpg.ifi.uzh.ch/ultimateslam.html.
J Melia - One of the best experts on this subject based on the ideXlab platform.
-
an optimum interpolation method applied to the resampling of noaa avhrr data
IEEE Transactions on Geoscience and Remote Sensing, 1994Co-Authors: J F Moreno, J MeliaAbstract:Two main problems must be solved in the geometric processing of satellite data: geometric registration and resampling. When the data must be geometrically registered over a reference map, and particularly when the Output Pixel size is not the same as the original Pixel size, the quality of the resampling can determine the quality of the Output, not only in the visual appearance of the image, but also in the numerically interpolated values when used in multitemporal or multisensor studies. The "optimum" interpolation algorithm for AVHRR data is defined over a 6/spl times/6 window in order to: consider overlapping effects among adjacent Pixels. The response for each new Pixel R(x, y) is determined as a linear combination of the response R/sub i/(x/sub i/y/sub i/) of the surrounding Pixels in the window (i=1,36). The weighting coefficients /spl mu//sub i/ are calculated from the ground projection of the effective spatial response function for each AVHRR Pixel, taking into account the particular viewing angle and geometry of the Pixels on the ground. This method is intended to give an optimal interpolation of AVHRR scenes along all the scanline, in order to compensate for off-nadir radiometric alterations associated to the varying spatial resolution and the blurring introduced by the Pixel overlaps. The optimum method, as mathematically defined, is highly expensive in CPU time. Then, a big effort is necessary to implement the algorithms so that they could be operationally applied. Two approaches are considered: a general numerical method and a pseudo-analytical approximation. A Landsat TM image corresponding to the same date of the AVHRR image is used to test the quality of the radiometric interpolation procedure. >
Antonio Rosinol Vidal - One of the best experts on this subject based on the ideXlab platform.
-
ultimate slam combining events images and imu for robust visual slam in hdr and high speed scenarios
International Conference on Robotics and Automation, 2018Co-Authors: Antonio Rosinol Vidal, Henri Rebecq, Timo Horstschaefer, Davide ScaramuzzaAbstract:Event cameras are bioinspired vision sensors that Output Pixel-level brightness changes instead of standard intensity frames. These cameras do not suffer from motion blur and have a very high dynamic range, which enables them to provide reliable visual information during high-speed motions or in scenes characterized by high dynamic range. However, event cameras Output only little information when the amount of motion is limited, such as in the case of almost still motion. Conversely, standard cameras provide instant and rich information about the environment most of the time (in low-speed and good lighting scenarios), but they fail severely in case of fast motions, or difficult lighting such as high dynamic range or low light scenes. In this letter, we present the first state estimation pipeline that leverages the complementary advantages of these two sensors by fusing in a tightly coupled manner events, standard frames, and inertial measurements. We show on the publicly available Event Camera Dataset that our hybrid pipeline leads to an accuracy improvement of 130% over event-only pipelines, and 85% over standard-frames-only visual-inertial systems, while still being computationally tractable. Furthermore, we use our pipeline to demonstrate-to the best of our knowledge-the first autonomous quadrotor flight using an event camera for state estimation, unlocking flight scenarios that were not reachable with traditional visual-inertial odometry, such as low-light environments and high dynamic range scenes. Videos of the experiments: http://rpg.ifi.uzh.ch/ultimateslam.html.
Z Malenovsky - One of the best experts on this subject based on the ideXlab platform.
-
influence of woody elements of a norway spruce canopy on nadir reflectance simulated by the dart model at very high spatial resolution
Remote Sensing of Environment, 2008Co-Authors: Z Malenovsky, E Martin, L Homolova, J P Gastelluetchegorry, R Zuritamilla, Michael E Schaepman, Radek Pokorny, J G P W Clevers, Pavel CudlinAbstract:Abstract A detailed sensitivity analysis investigating the effect of woody elements introduced into the Discrete Anisotropic Radiative Transfer (DART) model on the nadir bidirectional reflectance factor (BRF) for a simulated Norway spruce canopy was performed at a very high spatial resolution (modelling resolution 0.2 m, Output Pixel size 0.4 m). We used such a high resolution to be able to parameterize DART in an appropriate way and subsequently to gain detailed understanding of the influence of woody elements contributing to the radiative transfer within heterogeneous canopies. Three scenarios were studied by modelling the Norway spruce canopy as being composed of i) leaves, ii) leaves, trunks and first order branches, and finally iii) leaves, trunks, first order branches and small woody twigs simulated using mixed cells (i.e. cells approximated as composition of leaves and/or twigs turbid medium, and large woody constituents). The simulation of each scenario was performed for 10 different canopy closures (CC = 50–95%, in steps of 5%), 25 leaf area index (LAI = 3.0–15.0 m2 m− 2, in steps of 0.5 m2 m− 2), and in four spectral bands (centred at 559, 671, 727, and 783 nm, with a FWHM of 10 nm). The influence of woody elements was evaluated separately for both, sunlit and shaded parts of the simulated forest canopy, respectively. The DART results were verified by quantifying the simulated nadir BRF of each scenario with measured Airborne Imaging Spectroradiometer (AISA) Eagle data (Pixel size of 0.4 m). These imaging spectrometer data were acquired over the same Norway spruce stand that was used to parameterise the DART model. The Norway spruce canopy modelled using the DART model consisted of foliage as well as foliage including robust woody constituents (i.e. trunks and branches). All results showed similar nadir BRF for the simulated wavelengths. The incorporation of small woody parts in DART caused the canopy reflectance to decrease about 4% in the near-infrared (NIR), 2% in the red edge (RE) and less than 1% in the green band. The canopy BRF of the red band increased by about 2%. Subsequently, the sensitivity on accounting for woody elements for two spectral vegetation indices, the normalized difference vegetation index (NDVI) and the angular vegetation index (AVI), was evaluated. Finally, we conclude on the importance of including woody elements in radiative transfer based approaches and discuss the applicability of the vegetation indices as well as the physically based inversion approaches to retrieve the forest canopy LAI at very high spatial resolution.