Correction to: EKLT: Asynchronous Photometric Feature Tracking Using Events and Frames

International Journal of Computer Vision - The original version of this article was unfortunately omitted to publish the footnote “The best result per row is highlighted in bold” in...     

Improving sub-pixel accuracy for long range stereo

Dense stereo algorithms are able to estimate disparities at all pixels including untextured regions. Typically these disparities are evaluated at integer disparity steps. A subsequent sub-pixel interpolation often fails to propagate smoothness constraints on a sub-pixel level.We propose to increase the sub-pixel accuracy in low-textured regions in four possible ways: First, we present an analysis that shows the benefit of evaluating the disparity space at fractional disparities. Second, we introduce a new disparity smoothing algorithm that preserves depth discontinuities and enforces smoothness on a sub-pixel level. Third, we present a novel stereo constraint (gravitational constraint) that assumes sorted disparity values in vertical direction and guides global algorithms to reduce false matches, especially in low-textured regions. Finally, we show how image sequence analysis improves stereo accuracy without explicitly performing tracking. Our goal in this work is to obtain an accurate 3D reconstruction. Large-scale 3D reconstruction will benefit heavily from these sub-pixel refinements.Results based on semi-global matching, obtained with the above mentioned algorithmic extensions are shown for the Middlebury stereo ground truth data sets. The presented improvements, called ImproveSubPix, turn out to be one of the top-performing algorithms when evaluating the set on a sub-pixel level while being computationally efficient. Additional results are presented for urban scenes. The four improvements are independent of the underlying type of stereo algorithm.     

AlphaPilot: autonomous drone racing

Autonomous Robots - This paper presents a novel system for autonomous, vision-based drone racing combining learned data abstraction, nonlinear filtering, and time-optimal trajectory planning. The...     

Modeling of the performance of high-power diode amplifier systemswith an optothermal microscopic spatio-temporal theory

We present a detailed theoretical analysis of the dependence of spatio-temporal carrier and light field dynamics of high-power diode amplifier systems on their geometry and facet reflectivities as well as on the spatial and spectral characteristics of the optical input beam. The basis of the numerical modeling is the Maxwell-Bloch equations for spatially inhomogeneous semiconductor lasers which are self-consistently coupled to the nonequilibrium temperature dynamics of the electron-hole plasma. They microscopically describe the interaction between the optical fields, the charge carriers, and the interband polarization. Our numerical modeling allows an identification of the influence of dynamic internal laser effects such as diffraction, self-focusing, scattering, carrier diffusion, and heating on the performance of broad-area or tapered amplifiers (e.g., far field, near field). It thus provides a means of optimizing the epitaxial structure and geometry of high-power diode amplifier systems     

Trace metals in ambient air: Hourly size-segregated mass concentrations determined by synchrotron-XRF

Trace metals are ubiquitous in urban ambient air, with mass concentrations in the range of a few microg/m3 down to less than 100 pg/m3. To measure such low concentrations represents a challenge for chemical and physical analysis. In this study, ambient aerosol was collected in Zürich (Switzerland) in 1-h intervals and three size fractions (aerodynamic diameters 0.1-1 microm, 1-2.5 microm, and 2.5-10 microm), using a three-stage rotating drum impactor (RDI). The samples were analyzed by energy-dispersive Synchrotron radiation X-ray fluorescence spectrometry (SR-XRF) to obtain size-segregated hourly elemental aerosol mass concentrations for Cr, Mn, Fe, Cu, Zn, Br, and Pb, along with S, Cl, and Ca under the selected experimental conditions. The high sensitivity of SR-XRF allowed for detection limits of <50 pg/m3 for most of the above elements, with a net analysis time of only 15 s per sample. The data obtained with this technique illustrate that there is a considerable gain of relevant information when time resolution for measurements is increased from 1 day to 1 h. The individual size fractions of a specific element may show significantly different short-term patterns.     

Real-world emission factors of fine and ultrafine aerosol particles for different traffic situations in Switzerland

Extended field measurements of particle number (size distribution of particle diameters, D, in the range between 18 nm and 10 microm), surface area concentrations, and PM1 and PM10 mass concentrations were performed in Switzerland to determine traffic emissions using a comprehensive set of instruments. Measurements took place at roads with representative traffic regimes: at the kerbside of a motorway (120 km h(-1)), a highway (80-100 km h(-1)), and in an urban area with stop-and-go traffic (0-50 km h(-1)) regulated by light signals. Mean diurnal variations showed that the highest pollutant concentrations were during the morning rush hours, especially of the number density in the nanoparticle size range (D <50 nm). From the differences between up- and downwind concentrations (or differences between kerbside and background concentrations for the urban site), "real-life" emission factors were derived using NOx concentrations to calculate dilution factors. Particle number and volume emission factors of different size ranges (18-50 nm, 18-100 nm, and 18-300 nm) were derived for the total vehicle fleet and separated into a light-duty (LDV) and a heavy-duty vehicle (HDV) contribution. The total particle number emissions per vehicle were found to be about 11.7-13.5 x 10(14) particles km(-1) for constant speed (80-120 km h(-1) and 3.9 x 10(14) particles km(-1) for urban driving conditions. LDVs showed higher emission factors at constant high speed than under urban disturbed traffic flow. In contrast, HDVs emitted more air pollutants during deceleration and acceleration processes in stop-and-go traffic than with constant speed of about 80 km h(-1). On average, one HDV emits a 10-30 times higher amount of particulate air pollutants (in terms of both number and volume) than one LDV.     

EKLT: Asynchronous Photometric Feature Tracking Using Events and Frames

International Journal of Computer Vision - We present EKLT, a feature tracking method that leverages the complementarity of event cameras and standard cameras to track visual features with high...