Theses and Dissertations
This collection includes both ASU Theses and Dissertations, submitted by graduate students, and the Barrett, Honors College theses submitted by undergraduate students.
Displaying 1 - 2 of 2
Filtering by
- Creators: Gao, Tianxiang
- Creators: Turaga, Pavan
Description
Semantic image segmentation has been a key topic in applications involving image processing and computer vision. Owing to the success and continuous research in the field of deep learning, there have been plenty of deep learning-based segmentation architectures that have been designed for various tasks. In this thesis, deep-learning architectures for a specific application in material science; namely the segmentation process for the non-destructive study of the microstructure of Aluminum Alloy AA 7075 have been developed. This process requires the use of various imaging tools and methodologies to obtain the ground-truth information. The image dataset obtained using Transmission X-ray microscopy (TXM) consists of raw 2D image specimens captured from the projections at every beam scan. The segmented 2D ground-truth images are obtained by applying reconstruction and filtering algorithms before using a scientific visualization tool for segmentation. These images represent the corrosive behavior caused by the precipitates and inclusions particles on the Aluminum AA 7075 alloy. The study of the tools that work best for X-ray microscopy-based imaging is still in its early stages.
In this thesis, the underlying concepts behind Convolutional Neural Networks (CNNs) and state-of-the-art Semantic Segmentation architectures have been discussed in detail. The data generation and pre-processing process applied to the AA 7075 Data have also been described, along with the experimentation methodologies performed on the baseline and four other state-of-the-art Segmentation architectures that predict the segmented boundaries from the raw 2D images. A performance analysis based on various factors to decide the best techniques and tools to apply Semantic image segmentation for X-ray microscopy-based imaging was also conducted.
In this thesis, the underlying concepts behind Convolutional Neural Networks (CNNs) and state-of-the-art Semantic Segmentation architectures have been discussed in detail. The data generation and pre-processing process applied to the AA 7075 Data have also been described, along with the experimentation methodologies performed on the baseline and four other state-of-the-art Segmentation architectures that predict the segmented boundaries from the raw 2D images. A performance analysis based on various factors to decide the best techniques and tools to apply Semantic image segmentation for X-ray microscopy-based imaging was also conducted.
ContributorsBarboza, Daniel (Author) / Turaga, Pavan (Thesis advisor) / Chawla, Nikhilesh (Committee member) / Jayasuriya, Suren (Committee member) / Arizona State University (Publisher)
Created2020
Description
Single drop impact of liquid on a static powder bed was studied to investigate the granule formation mechanism, droplet penetration time, the characterization of granules (morphology, surface structure and internal structure), as well as the formation regime map. Water was used as the liquid and two pharmaceutical powders, microcrystalline cellulose (MCC) and acetaminophen (APAP), were mixed to make heterogeneous powder beds. The complete drop impact and penetration was recorded by a high-speed camera. Two granule formation mechanisms identified previously occurred: Spreading and Tunneling. Spreading occurred for mixtures of large particle sizes, while Tunneling started to occur when the particle sizes of the mixtures decreased. With an increase of APAP concentration, the overall drop penetration time increased, which was in good agreement with previous literature. The granule morphology, surface structure, and internal structure were characterized by a prism method with image analysis, scanning electron microscope, and X-ray microtomography, respectively. The Spreading mechanism produced flat disks with porous internal structures, while the Tunneling mechanism produced round granules with dense internal structures. Granules that were formed via a hybrid of the mechanisms, Spreading/Tunneling, were hybrid granules, with some dense areas and some porous areas. The results of the granule content uniformity from UV-vis spectrometry revealed that with the increase of APAP proportion, the overall uniformity was compromised for mixtures with fine ingredients, while the content was much more uniform for coarse mixtures. It is believed that the mean particle size of the powder bed is the predominant factor in influencing the formation mechanism, drop penetration time, and granule properties, while the content uniformity is affected by both the particle sizes and the mixture hydrophobicity.
ContributorsGao, Tianxiang (Author) / Emady, Heather N (Thesis advisor) / Chawla, Nikhilesh (Committee member) / Jiao, Yang (Committee member) / Pradhan, Shankali (Committee member) / Oka, Sarang (Committee member) / Arizona State University (Publisher)
Created2020