In this article, we introduce a parallel algorithm for connected-component analysis (CCA) on GPUs which drastically reduces the volume of data to transfer from GPU to the host. CCA algorithms targeting GPUs typically store the extracted features in arrays large enough to potentially hold the maximum possible number of objects for the given image size. Transferring these large arrays to the host requires large portions of the overall execution time. Therefore, we propose an algorithm which uses a CUDA kernel to merge trees of connected component feature structs. During the tree merging, various connected-component properties, such as total area, centroid and bounding box, are extracted and accumulated. The tree structure then enables us to only transfer features of valid objects to the host for further processing or storing. Our benchmarks show that this implementation significantly reduces memory transfer volume for processing results on the host whilst maintaining similar performance to state-of-the-art CCA algorithms.
1 July 2019 in Transactions of the Institute of Measurement and Control
In this paper, a novel approach for tracking moving structures in multiphase flows over larger axial ranges is presented, which at the same time allows imaging the tracked structures and their environment. For this purpose, ultrafast electron beam X-ray computed tomography (UFXCT) is being extended by an image-based position control. Application is scanning and tracking of, for example, bubbles, particles, waves and other features of multiphase flows within vessels and pipes. Therefore, the scanner has to be automatically traversed with the moving structure basing on real-time scanning, image reconstruction and image data processing. In this paper, requirements and different strategies for reliable object tracking in dual image plane imaging mode are discussed. Promising tracking strategies have been numerically implemented and evaluated.
27 October 2021 in IEEE Transactions on Nuclear Science
Ultrafast X-ray computed tomography (UFXCT) is a fast tomographic imaging technique based on the principle of electron beam scanning. It is used for the investigation of transient multiphase flows. A UFXCT scanner comprises multiple detector modules generating gigabytes of raw data per second for imaging rates of up to 8000 fps. During the data acquisition, these data are stored in RAM on each detector module and can therefore only be accessed after the acquisition. This limits the applicability of UFXCT systems because real-time control of the scanned process or the scanner itself is not possible. To achieve this, data streaming with low latency and high throughput is necessary. Using the newly developed FPGA-based data acquisition described in this article, the data acquisition is enhanced to real-time operation, i.e., streaming the acquired data to the processing computer within microseconds of their acquisition. The developed FPGA hardware implementation handles control of peripherals [such as analog-digital converters (ADC)], synchronization between detector modules, electron beam deflection, and data transfer via Gigabit Ethernet (GbE).
In this paper, a smart detector design for novel multi-plane ultrafast electron beam X-ray computed tomography is presented. The concept is based on multi-plane electron beam scanning on a transparent X-ray target and elongated cuboid-shape scintillation detectors for radiation detection over an extended axial scanning range. The optical part of the scintillation detector acts as both an X-ray sensitive scintillator with a fast time response and a light guide. With that, we reduce detector complexity, number of detector elements, overall power consumption, and detector costs. We investigated the performance of this new multi-plane detector design with an evaluation detector setup that is made of cerium doped lutetium yttrium oxyorthosilicate (LYSO:Ce) as scintillation material and an avalanche photodiode (APD) array. Thereby, we assessed two design variants: A monolithic LYSO bar detector and a sandwich detector made of multiple LYSO crystals and glass light-guides. Both types reveal excellent linear detector responses, long-term stabilities, and comparable signal qualities.
Ultrafast X-ray computed tomography is an advanced imaging technique for multiphase flows. It has been used with great success for studying gas–liquid as well as gas–solid flows. Here, we apply this technique to analyze density-driven particle segregation in a rotating drum as an exemplary use case for analyzing industrial particle mixing systems. As glass particles are used as the denser of two granular species to be mixed, beam hardening artefacts occur and hamper the data analysis. In the general case of a distribution of arbitrary materials, the inverse problem of image reconstruction with energy-dependent attenuation is often ill-posed. Consequently, commonly known beam hardening correction algorithms are often quite complex. In our case, however, the number of materials is limited. We therefore propose a correction algorithm simplified by taking advantage of the known material properties, and demonstrate its ability to improve image quality and subsequent analyses significantly.