A framework for estimating relative depth in video

This dissertation proposes a novel framework for recovering relative depth maps from a video. The framework is composed of two parts: a depth estimator and a sparse label interpolator. These parts are completely separate from one another and can operate independently. Prior methods have tended to heavily couple the interpolation stage with the depth estimation, which can assist with automation at the expense of flexibility. The loss of this flexibility can in fact be worse than any advantage gained by coupling the two stages together. This dissertation shows how by treating the two stages separately, it is very easy to change the quality of the results with little effort. It also leaves room for other adjustments. The depth estimator is based upon well-established computer vision principles and only has the restriction that the camera must be moving in order to obtain depth estimates. By starting from first principles, this dissertation has developed a new approach for quickly estimating relative depth. That is, it is able to answer the question, “is this feature closer than another," with relatively little computational overhead. The estimator is designed using a pipeline-style approach so that it produces sparse depth estimates in an online fashion; i.e. a depth estimate is automatically available for each new frame presented to the estimator. Finally, the interpolator applies an existing method based upon edge-aware filtering to generate the final depth maps. When temporal filters are used, the interpolation stage is able to very easily handle frames without any depth information, such as when the camera was stationary. However, unlike the prior work, this dissertation establishes the theoretical background for this type of interpolation and addresses some of the associated numerical problems. Strategies for dealing with these issues have also been provided

A framework for estimating relative depth in video

This dissertation proposes a novel framework for recovering relative depth maps from a video. The framework is composed of two parts: a depth estimator and a sparse label interpolator. These parts are completely separate from one another and can operate independently. Prior methods have tended to heavily couple the interpolation stage with the depth estimation, which can assist with automation at the expense of flexibility. The loss of this flexibility can in fact be worse than any advantage gained by coupling the two stages together. This dissertation shows how by treating the two stages separately, it is very easy to change the quality of the results with little effort. It also leaves room for other adjustments. The depth estimator is based upon well-established computer vision principles and only has the restriction that the camera must be moving in order to obtain depth estimates. By starting from first principles, this dissertation has developed a new approach for quickly estimating relative depth. That is, it is able to answer the question, “is this feature closer than another," with relatively little computational overhead. The estimator is designed using a pipeline-style approach so that it produces sparse depth estimates in an online fashion; i.e. a depth estimate is automatically available for each new frame presented to the estimator. Finally, the interpolator applies an existing method based upon edge-aware filtering to generate the final depth maps. When temporal filters are used, the interpolation stage is able to very easily handle frames without any depth information, such as when the camera was stationary. However, unlike the prior work, this dissertation establishes the theoretical background for this type of interpolation and addresses some of the associated numerical problems. Strategies for dealing with these issues have also been provided

The Use of Euler Deconvolution Technique in the Estimation of Depth to Magnetic Source Bodies around the Schist Belt Parts of Kano State, Nigeria

Depth estimation of magnetic source bodies in parts of the Schist Belt of Kano, using Euler Deconvolution is presented in this paper. Detail ground magnetic survey was carried out using SCINTREX proton precession magnetometer to produce the Total Magnetic Intensity (TMI) map and consequently the residual map. The TMI ranges from 34,261 nT to 34,365 nT, while the residual field ranges from -160 nT to 115 nT. The depth estimate for contacts ranges from 6.5 m to 39.8 m, while that of dyke ranges from 8.9 m to 51.3 m. The depth estimation presented in this work is compared with the results of aeromagnetic study carried out in the same area and found to agree fairly well. Further, this also ensures the validity of aeromagnetic investigation in such applications. Keywords: Contacts, Dykes, Euler Deconvolution, Schist Belt. PACS: 91.25.F and 91.25.Rt.

Temporal Variations in the Depth Estimate of Mass Density Changes at Mt Etna Volcano

<p>Continuous gravity measurements at Mt. Etna, Sicily demonstrate spatio-temporal variations that can be related to volcanic processes. Two iGrav superconducting gravimeters (SGs) were installed in 2014 and 2016 at Serra La Nave Astrophysical Observatory (SLN; 1,730 m elevation; ~6.5 km from the summit craters) and La Montagnola hut (MNT; 2,600 m asl; ~3.5 km SE of the summit crater). Since their installation both stations have been continuously recording, resulting in high-resolution (1 Hz sampling rate) time series. The persistent activity of Etna is maintained by a regular supply of magma to the shallower levels of the plumbing system. The bulk mass redistributions induced by the newly injected material result in minor variations in the local gravity field that are recorded by the two stations. By assuming that the observed gravity changes are due exclusively to mass changes in an almost spherical-shaped source, located beneath the craters, the amplitude ratio between the two signals can be used to estimate the depth to potential mass changes beneath the surface.</p><p>This study reports on the time-dependent nature of mass changes located beneath the craters of the volcano during 2019. Results highlight distinct periods of stability at different depths, as well as potential periods of transitory activity, where the predominant mass source switches between storage zones at different depth. These events are correlated to phases of strombolian and effusive activity, highlighting the potential of continuous gravimetry for the detection of eruption precursors.</p>

CNN Based Monocular Depth Estimation

In several applications, such as scene interpretation and reconstruction, precise depth measurement from images is a significant challenge. Current depth estimate techniques frequently provide fuzzy, low-resolution estimates. With the use of transfer learning, this research executes a convolutional neural network for generating a high-resolution depth map from a single RGB image. With a typical encoder-decoder architecture, when initializing the encoder, we use features extracted from high-performing pre-trained networks, as well as augmentation and training procedures that lead to more accurate outcomes. We demonstrate how, even with a very basic decoder, our approach can provide complete high-resolution depth maps. A wide number of deep learning approaches have recently been presented, and they have showed significant promise in dealing with the classical ill-posed issue. The studies are carried out using KITTI and NYU Depth v2, two widely utilized public datasets. We also examine the errors created by various models in order to expose the shortcomings of present approaches which accomplishes viable performance on KITTI besides NYU Depth v2.

Observations on the May 2019 Joffre Peak landslides, British Columbia

Two catastrophic landslides occurred in quick succession on 13 and 16 May 2019, from the north face of Joffre Peak, Cerise Creek, southern Coast Mountains, British Columbia. With headscarps at 2560 m and 2690 m elevation, both began as rock avalanches, rapidly transforming into debris flows along middle Cerise Creek, and finally into debris floods affecting the fan. Beyond the fan margin, a flood surge on Cayoosh Creek reached bankfull and attenuated rapidly downstream; only fine sediment reached Duffey Lake. The toe of the main debris flow deposit reached 4 km from the headscarp, with a travel angle of 0.28, while the debris flood phase reached the fan margin 5.9 km downstream, with a travel angle of 0.22. Photogrammetry indicates the source volume of each event is 2–3 Mm3, with combined volume of 5 Mm3. Lidar differencing, used to assess deposit volume, yielded a similar total result, although error in the depth estimate introduced large volume error masking the expected increase due to dilation and entrainment. The average velocity of the rock avalanche-debris flow phases, from seismic analysis, was ~ 25–30 m/s, and the velocity of the 16 May debris flood on the upper fan, from super-elevation and boulder sizes, was 5–10 m/s. The volume of debris deposited on the fan was ~ 104 m3, 2 orders of magnitude less than the avalanche/debris flow phases. Progressive glacier retreat and permafrost degradation were likely the conditioning factors; precursor rockfall activity was noted at least ~6 months previous; thus, the mountain was primed to fail. The 13 May landslide was apparently triggered by rapid snowmelt, with debuttressing triggering the 16 May event.

An Accelerated Cue Combination Principle Accounts for Multi-cue Depth Perception

For the visual world in which we operate, the core issue is to conceptualize how its three-dimensional structure is encoded through the neural computation of multiple depth cues and their integration to a unitary depth structure. One approach to this issue is the full Bayesian model of scene understanding, but this is shown to require selection from the implausibly large number of possible scenes. An alternative approach is to propagate the implied depth structure solution for the scene through the “belief propagation” algorithm on general probability distributions. However, a more efficient model of local slant propagation is developed as an alternative.The overall depth percept must be derived from the combination of all available depth cues, but a simple linear summation rule across, say, a dozen different depth cues, would massively overestimate the perceived depth in the scene in cases where each cue alone provides a close-to-veridical depth estimate. On the other hand, a Bayesian averaging or “modified weak fusion” model for depth cue combination does not provide for the observed enhancement of perceived depth from weak depth cues. Thus, the current models do not account for the empirical properties of perceived depth from multiple depth cues.The present analysis shows that these problems can be addressed by an asymptotic, or hyperbolic Minkowski, approach to cue combination. With appropriate parameters, this first-order rule gives strong summation for a few depth cues, but the effect of an increasing number of cues beyond that remains too weak to account for the available degree of perceived depth magnitude. Finally, an accelerated asymptotic rule is proposed to match the empirical strength of perceived depth as measured, with appropriate behavior for any number of depth cues.