The state of the art of wave field synthesis (WFS) systems is that they can reproduce sound sources and secondary (mirror image) sources with natural spaciousness in a horizontal plane, and thus perform satisfactory 2D auralization of an enclosed space, based on multitrace impulse response data measured or simulated along a 2D microphone array. However, waves propagating with a nonzero elevation angle are also reproduced in the horizontal plane, which is neither physically nor perceptually correct. In most listening environments to be auralized, the floor is highly absorptive since it is covered with upholstered seats, occupied during performances by a well-dressed audience. A first-order ceiling reflection, reaching the floor directly or via a wall, will be severely damped and will not play a significant role in the room response anymore. This means that a spatially correct WFS reproduction of first-order ceiling reflections, by means of a loudspeaker array at the ceiling of the auralization reproduction room, is necessary and probably sufficient to create the desired 3D spatial perception. To determine the driving signals for the loudspeakers in the ceiling array, it is necessary to identify the relevant ceiling reflection(s) in the multichannel impulse response data and separate those events from the data set. Two methods are examined to identify, separate, and reproduce the relevant reflections: application of the Radon transform, and decomposition of the data into cylindrical harmonics. Application to synthesized and measured data shows that both methods in principle are able to identify, separate, and reproduce the relevant events.

In many ultrasonic imaging systems, data acquisition and image formation are performed on separate computing devices. Data transmission is becoming a bottleneck, thus, efficient data compression is essential. Compression rates can be improved by considering the fact that many image formation methods rely on approximations of wave-matter interactions, and only use the corresponding part of the data. Tailored data compression could exploit this, but extracting the useful part of the data efficiently is not always trivial. In this work, we tackle this problem using deep neural networks, optimized to preserve the image quality of a particular image formation method. The Delay-And-Sum (DAS) algorithm is examined which is used in reflectivity-based ultrasonic imaging. We propose a novel encoder-decoder architecture with vector quantization and formulate image formation as a network layer for endto-end training. Experiments demonstrate that our proposed data compression tailored for a specific image formation method obtains significantly better results as opposed to compression agnostic to subsequent imaging. We maintain high image quality at much higher compression rates than the theoretical lossless compression rate derived from the rank of the linear imaging operator. This demonstrates the great potential of deep ultrasonic data compression tailored for a specific image formation method.

Ultrasonic imaging algorithms used in many clinical and industrial applications consist of three steps: A data preprocessing, an image formation and an image post-processing step. For efficiency, image formation often relies on an approximation of the underlying wave physics. A prominent example is the Delay-And-Sum (DAS) algorithm used in reflectivity-based ultrasonic imaging. Recently, deep neural networks (DNNs) are being used for the data pre-processing and the image postprocessing steps separately. In this work, we propose a novel deep learning architecture that integrates all three steps to enable endto-end training. We examine turning the DAS image formation method into a network layer that connects data pre-processing layers with image post-processing layers that perform segmentation. We demonstrate that this integrated approach clearly outperforms sequential approaches that are trained separately. While network training and evaluation is performed only on simulated data, we also showcase the potential of our approach on real data from a non-destructive testing scenario.

Articles you may be interested inAutomated flaw detection scheme for cast austenitic stainless stell weld specimens using Hilbert-Huang transform of ultrasonic phased array data AIP Conf.Quantitative evaluation of ultrasonic sound fields in anisotropic austenitic welds using 2D ray tracing model AIP Conf.ABSTRACT. Ultrasonic imaging can be applied for the inspection of welds to determine the type, position, and size of defects. By including boundary reflections in the path from transmitter to receiver, defects in the weld can be imaged from different directions using both reflection and diffraction signals. These images can be combined into a single figure providing a section view of the weld. The geometrical configuration must be taken into account for proper alignment of the subimages. These sub-images can be linked to conventional ultrasonic inspection techniques, such as the phased array tandem technique and time of flight diffraction.

No abstract