DSB Seminars 2016

Shear wave attenuation and micro-fluidics in water-saturated sand and glass beads. An improvement in the modeling of shear wave attenuation and speed in water-saturated sand and glass beads is introduced. Some dry and water-saturated materials are known to follow a constant-Q model in which the attenuation, expressed as Q⁻¹, is independent of frequency. The associated loss mechanism is thought to lie within the solid frame. A second loss mechanism in fluid-saturated porous materials is the viscous loss due to relative motion between pore fluid and solid frame predicted by the Biot–Stoll model. It contains a relaxation process that makes the Q⁻¹ change with frequency, reaching a peak at a characteristic frequency. Examination of the published measurements above 1 kHz, particularly those of Brunson (Ph.D. thesis, Oregon State University, Corvalis, 1983), shows another peak, which is explained in terms of a relaxation process associated with the squirt flow process at the grain-grain contact. In the process of deriving a model for this phenomenon, it is necessary to consider the micro-fluidic effects associated with the flow within a thin film of water confined in the gap at the grain-grain contact and the resulting increase in the effective viscosity of water. The result is an extended Biot model that is applicable over a broad band of frequencies.. See also JASA 2014

Modelling biomedical ultrasound. Modelling the propagation of ultrasound waves in the human body has many applications, from reconstructing images to planning treatments using ultrasound therapy. In this talk, I will discuss the steps required in developing a numerical model, including forming equations that capture the physics of interest, applying appropriate numerical methods to solve these equations, developing computer codes for different computer architectures, and performing experimental validation. To create useful models, each of these steps and the corresponding constraints must be carefully considered. Using these steps as a blue-print, I will then discuss the development of k-Wave, a MATLAB toolbox that we have developed for modelling ultrasound waves in the body.

Kaja Kvåle,

Ole Marius Hoel Rindal

Kaja Kvåle : Summary of the IEEE IUS 2016

Ole Marius Hoel Rindal : Summary of the PICMUS IEEE IUS 2016 challenge

Title: Deep neural networks - a powerful tool to solve your computer vision problems

Abstract: After 2012, when deep learning based techniques won the ImageNet contest with a clear margin to competing algorithms, deep neural networks have revolutionized computer vision. For many classification tasks deep learning has drastically surpassed previous state of the art results in classification accuracy. Currently large deep neural networks achieve the best results on speech recognition, visual object recognition, character recognition, and several language related tasks. Deeper machine learning architectures are better capable of handling complex recognition tasks compared to previous more shallow models.
Major benefits of deep networks include:

(i) their superior modelling capabilities of heterogeneous data in layers of increasing complexity,
(ii) their ability to learn the best features to represent the raw data, and
(iii) their ability to gain in performance with the availability of more training data.

In this talk we will give a basic introduction to deep learning, aiming to describe typical designs and learning strategies. We show some challenging computer vision applications where we at NR have applied deep learning with great success.

Acoustic radiation force - optical coherence elastography

27. May

(Friday 11.15-12)

Sverre
Holm

Lack of coherence between planetary and climate oscillations and an unresolved radio mystery

NB! Astrophysics Department: Rom 304 (Peisestua), Institutt for teoretisk astrofysikk

Trond Bergh,

Elsa Cecconello, 

and Anders Ueland

Trond Bergh: Multi-Speaker Voice Activity Detection Using a Camera-assisted Microphone Array

Elsa Cecconello: Modelling seismic data for time-varying rough sea surfaces

Anders Ueland : 3D attributes and classification of salt bodies in unlabelled datasets.

Connecting the Viscous Grain-shearing Mechanism of Wave Propagation in Marine Sediments to Fractional Calculus

Fabrice Prieur

Title: Reflections on Ocean Bottom Reflections: a perspective about inversions

NB! Tuesday 13.15-14.00, seminar room Logo, 2438

Abstract: The sound field measured at a receiver in the ocean contains information about the physical characteristics of the ocean environment. This paper focuses on geoacoustic inversion, and describes specific cases in which experiments with sound sources have been used to learn about the physical properties and the structure of the ocean bottom, or to understand the physical processes at work in the ocean bottom that generate the features we observe.  The formalism of Bayesian inference is reviewed briefly to establish an understanding of the approach that is in widespread use.  Then two recent examples of applications in acoustical oceanography are discussed. 

The first is a Bayesian inversion of ocean bottom reflection coefficient versus angle data to estimate geoacoustic model parameters of young oceanic crust at deep water sites in the Pacific Ocean.  The experimental data were obtained in an experiment to study the variation of sound speed in crustal basalt with age of the crust. 

The second example is from a tomography experiment using moored ocean bottom hydrophones to study the sound speed structure at a gas hydrate vent site in the Cascadia Margin off the west coast of British Columbia.  The inversion is an example of a linearized approach that uses travel time information of bottom and sub-bottom reflected signals.

Title: Rehearsal for invited presentations at AIUM meeting in New York 17-21 March

Andreas: "Adaptive Beamforming for High-Resolution Imaging" (17 min) in session "Synthetic Focus Imaging Techniques"

Sverre: "Processing the data twice: Minimum variance as an alternative to geometry-based beamforming" (25 min) in session "Cleaning Up Ultrasound Images"

Title: SURF (Second Order UltRasound Field) imaging.

NB! Monday14.15-14.45