Master of Science (MS), Wright State University, 2020, Computer Science

This paper presents a framework for the generation of 3D models. This is an important problem for many reasons. For example, 3D models are important for systems that are involved in target recognition. These systems use 3D models to train up accuracy on identifying real world object. Traditional means of gathering 3D models have limitations that the generation of 3D models can help overcome. The framework uses a novel generative adversarial network (GAN) that learns latent representations of two dimensional views of a model to bootstrap the network's ability to learn to generate three dimensional objects. The novel architecture is evaluated using two different types of evaluation. The two dimensional views are evaluated using a combination of an Inception Score and Hausdorff Distance, compared against the two dimensional views of the real 3D models used in training. The three dimensional object are evaluated using the Hausdorff Distance compared against the real 3D models used in training. Experimental results demonstrate that the novel generative adversarial network that is being proposed generated realistic looking models faster, and with higher fidelity than a basic 3D generative adversarial network would produce with the same training structure. The thesis illustrates the promise of GAN bootstrapping with two dimensional perspective codes to create higher fidelity three dimensional models.

Committee: Derek Doran Ph.D. (Advisor); John Gallagher Ph.D. (Committee Member); Fred Garber Ph.D. (Committee Member) Subjects: Computer Science

Master of Science (MS), Bowling Green State University, 2023, Geology

Ammonoid suture patterns have long been a point of interest in the study of this extinct cephalopod group, as they can be used for phylogenetic classification, understanding organismal growth, and many other topics of inquiry. These patterns are traditionally recorded as flat lines through tracings made using tape placed on fossil specimens, which causes distortion of the original shape. This problem prompts the need for a solution that may help to increase the accuracy of the data collected from suture patterns. Fourteen ammonoid specimens representing nine species and four suborders and spanning the Middle Devonian to Late Cretaceous Periods were digitally modeled, and their 3D digital suture patterns were then compared to 2D tracings. First, 3D models of the specimens were made using photogrammetry, which is a low-cost method with a low barrier to entry done by compiling photographs taken of a specimen and stitching them together into a 3D model using open-source software. The original goal was to then digitally flatten the 3D models using additional software to create 2D representations of their suture patterns that could then be compared to the tracings created using the traditional hand-traced tape method. While there were technical complications with the 3D to 2D conversion, the method of photogrammetry produced accurate depictions of the original specimens, based on visual comparison of the 3D and 2D suture patterns. Several notable factors influenced the quality of the 3D model created using this photogrammetry method, including the preserved definition of the suture pattern, the reflectiveness of the fossil's surface, and the uniformity of color of the fossil. With these factors taken into consideration, photogrammetry can be a very accurate and cost-effective method for digitizing fossils, opening up a range of potential new morphometric analyses and expanding access to these important paleontological specimens beyond those able to physically visit fossil (open full item for complete abstract)

Committee: Margaret Yacobucci Ph. D (Committee Chair); Yuning Fu Ph. D (Committee Member); Peter Gorsevski Ph. D (Committee Member) Subjects: Geology; Paleontology

Doctor of Philosophy, The Ohio State University, 2018, Civil Engineering

This research focuses on the wall and the towers of the Great Wall of China in Jinshanling, located in Luanping County, approximately 153 km northeast of Beijing, China. The study reveals that the main construction methods used to create the wall were rammed earth and rubble construction, foundation stone masonry installation, and mostly Flemish configuration brick masonry installation. These methods were implemented in a bottom-up construction. In some areas, especially in the eastern area of Jinshanling, foundation stones and bricks were integrated with existing stone masonry (large stones and mortar) from an earlier construction period at the beginning of the Ming Dynasty. The main construction method used to construct the towers was also rammed earth and rubble construction for the base of the tower's inner core of the base. Similarly, to the wall, the towers were completed with an outer layer at the base composed of fire kiln bricks and foundation stones. The installation of fire kiln brick masonry and foundation stone masonry were implemented on the outer layer of the towers' base. The main difference between the towers is the support system on the first floor. One design used timber columns on the first floor to support the second floor of the towers, while a second design used interior brick walls, arches, and vaults on the first floor to support the second floor. There is a statistical correlation showing that the bricks and arches supporting the towers were more likely to be implemented at a higher elevation in sampled towers. The findings from literature searches, site visits, and interviews of experts on the Great Wall were used as the inputs for the fuzzy sets and logic assessments. The fuzzy models used in the evaluations were the fuzzy sets angular model and the fuzzy sets rotational model. The outputs of the fuzzy evaluations (i.e., the main construction method, the sequence implemented at the time of construction, and the current performance of t (open full item for complete abstract)

Committee: Fabian Hadipriono Tan (Advisor) Subjects: Civil Engineering; History

Master of Science in Engineering, Youngstown State University, 2017, Department of Mechanical, Industrial and Manufacturing Engineering

Aircraft operators must maintain and sustain their aircraft through the platform's life cycle. The Department of Defense (DoD) is no exception. Many DoD missions may require a time-sensitive production of spare parts. This lends itself to spare parts production by the Department of Defense itself and such an approach could be enabled by additive manufacturing. In order for the government to be able to produce spare parts in-house an entirely new business model between the original equipment manufacturer (OEM) and the government has to be established. A physical spare part would not be the transacted item; instead the technical data package (TDP) would be exchanged. Industry needs to be incentivized to adopt a data focused business model. A key question is can industry achieve equivalent profit similarly to the traditional spare parts production? This research explores business models from the perspective of industry. A survey was provided to both government and industry to identify differences and similarities in assumptions and expectations. Four different business models were developed. The business models were applied to two different case studies to evaluate the pros and cons of the various models. This analysis provides industry and government a reference for discussions on approaches toward future maintenance and sustainment manufacturing operations.

Committee: Brett Conner PhD (Advisor); Darrell Wallace PhD (Committee Member); Martin Cala PhD (Committee Member) Subjects: Business Costs; Engineering; Industrial Engineering; Public Policy

Doctor of Philosophy (PhD), Wright State University, 2011, Computer Science and Engineering PhD

A method to automatically characterize and identify 3-D range scans based on intrinsic landmarks is presented. Intrinsic landmarks represent locally unique, intrinsic properties of a scanned surface, regardless of scale or rotation. The number, location, and characteristics of landmarks are used to characterize the scanned models. This method contains a selection process to identify stable, intrinsic landmarks for range scans as well as the identification of those scans. The selection process requires no user interaction or surface assumptions. It uses the principal curvatures at the range points to select the landmarks. First, a large number of landmarks are generated by fitting a bi-cubic polynomial surface to points surrounding each range point and calculating the principal curvatures at the range point. Points of locally extremum principal curvature are then considered candidate landmarks. Using a random sample and consensus (RANSAC) algorithm, candidate landmarks that match with landmarks in other scans of the same subject are selected as final, stable landmarks. Our main goal is to provide a means to characterize models in a range data base. With several scans of each subject available in the data base, a number of stable landmarks are determined for each subject. The locations and characteristics of the landmarks are used to describe a subject and distinguish it from other subjects. The main contribution of this work is considered to be the selection of unique and stable landmarks in a range scan and generation of a descriptor for each landmark that characterizes the intrinsic properties of the surface in the neighborhood of the landmark. The effectiveness of the method is presented through the successful identification of processed subjects and characterization of new subjects.

Committee: Arthur Goshtasby PhD (Committee Chair); Jack Jean PhD (Committee Member); Thomas Wischoll PhD (Committee Member); Lang Hong PhD (Committee Member); Kathleen Robinette PhD (Committee Member) Subjects: Computer Engineering

Doctor of Philosophy, The Ohio State University, 2009, Electrical and Computer Engineering

This dissertation presents a new set of three-dimensional scattering feature models for synthetic aperture radar (SAR). We develop a set of parametric models of canonical shapes that capture aspect-dependent, high-frequency scattering for bistatic (and monostatic) 3D SAR phase history responses. The models are parameterized by the shape location, orientation, and size as well as the radar transmitter and receiver antenna aspects and frequency. We develop the models by combining physical optics (PO) and uniform theory of diffraction (UTD) planar scattering solutions to approximate 3D scattering responses of canonical shapes. We validate the models using scattering prediction software and show that the proposed models capture well the mainlobe responses of each shape. Thus, the proposed models may be used to accurately predict first-order scattering of scenes comprised of such shapes.The second part of this dissertation focuses on the inverse problem of discerning the types of canonical shapes in a scene and estimating their corresponding model parameters from observed SAR phase history data. We present discrimination methods for classifying observed scattering into the geometric shape types. We compute the Cramer-Rao bounds for the models and characterize model parameter estimation accuracy for two estimation schemes. Finally, we present a feature extraction algorithm that classifies and estimates the canonical features from polarimetric phase history data. We use non-quadratic regularization to form sparsity-constrained 3D SAR images that are used to initialize the scatterer location, orientation, and size estimates. We test the feature extraction algorithm on simulated phase histories for densely-sampled and sparsely-sampled, monostatic and bistatic 3D SAR apertures. We show that even for sparsely-sampled apertures, the feature extraction algorithm is able to estimate geometric scattering features in the scene. Feature extraction for the proposed canonical shape mo (open full item for complete abstract)

Committee: Randolph Moses PhD (Advisor); Lee Potter PhD (Committee Member); Emre Ertin PhD (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2008, Electrical and Computer Engineering

A major goal in pattern recognition algorithms is to achieve theperformance of the Bayes optimal rule, i.e. minimum probability in classification error. Unfortunately, we usually can't achieve to this goal, because the original data distributions are unknown. This leaves us with the need to estimate the true, underlying class distributions from samples. This estimation procedure adds classification errors due to two major causes. First, the form of the density function used in our estimate may not correctly define the data. Second, noise and limited data available may generate incorrect estimates. In particular, the first problem usually occurs when the data representations share a common norm (spherical data). Since the estimation of the Gaussian model is much easier than those of spherical models, researchers generally resort to the uses of the former. In this thesis, we show that in some particular cases, which we named spherical-homoscedastic, one can use the Gaussian model and still obtain Bayes optimal classifications. We applied the developed theory to many practical problems including text classification, gene expression analysis and shape analysis. For the analysis of shapes, we introduce the new key concept of rotation invariant kernels. Here, we derive a criterion to select the parameter of this kernel that make the shape distributions spherical-homoscedastic in the kernel space. The second major problem is addressed by proposing a feature extraction algorithm considering the Bayes optimality of the solution. Similarly to the above classification problem, most of the algorithms defined to date are extracting the classification information depending on some discriminant criteria, rather than the Bayes error itself. This is due to the difficulties associated with calculating the Bayes error. In the second part of this thesis, we design an algorithm that can extract the 1-dimensional subspace where the Bayes error is minimized for homoscedastic (i.e., same c (open full item for complete abstract)

Committee: Aleix M. Martinez PhD (Advisor); Andrea Serrani PhD (Committee Member); Yoonkyung Lee PhD (Committee Member); Mikhail Belkin PhD (Committee Member) Subjects: Computer Science; Electrical Engineering; Statistics

Doctor of Philosophy, Miami University, 2013, Chemistry and Biochemistry

The abstract nature of bonding poses several challenges as some students struggle to identify the ionic and covalent nature of bonds within a variety of compounds. The topic is formally introduced in both middle school and secondary education (grades 6-12) and reintroduced in the undergraduate general chemistry curriculum. Teachers use multiple representations to communicate the concepts of bonding, including Lewis Structures, formulas, spacefilling models, and 3D manipulatives. As students learn to interpret these multiple representations, they may develop misconceptions that can create a problem in further learning of chemistry. Therefore, representations of chemical structures are useful to elicit student ideas of chemical bonding by interviewing students about the features of the models as they pertain to bonding. Students enrolled in high school physical science, high school chemistry, and general chemistry were interviewed using a five-phase semi-structured protocol focused on multiple representations to explore students' ideas about covalent and ionic bonding. Four themes emerged through constant comparative analysis for describing bond type: periodic trends, electrostatic interactions, octet rule, and representation surface features. The qualitative findings were used to develop the Bonding Representations Inventory (BRI) to determine the prevalence of the misconceptions. The BRI was administered to 1072 high school chemistry, advanced placement chemistry, and general chemistry students across the United States. Content validity was determined through expert validation of the items, and concurrent validity was established with the three groups of students. Reliability was determined through individual item analysis and through Ferguson's δ. The BRI can be used in high school and general chemistry classrooms to inform teaching of both bonding and representations. Additional research was conducted in inorganic chemistry education (Appendix A): Creating an In (open full item for complete abstract)

Committee: Stacey Lowery Bretz (Advisor); Ellen J. Yezerski (Committee Chair); Michael W. Crowder (Committee Member); David L. Tierney (Committee Member); Kathryn B. McGrew (Committee Member) Subjects: Chemistry; Education; Educational Tests and Measurements; Educational Theory; Higher Education; Inorganic Chemistry; Molecules; Science Education; Secondary Education; Teaching