No description available.

No description available.

No description available.

What information is used by the visual system to detect patterns? A standard model hypothesizes that both spatial frequency and orientation information are processed by independent channels, meaning there is no summation among channels. Despite consensus among researchers on how the visual system sums spatial frequency and orientation information there are data in the literature (Kersten, 1987) that ostensibly contradict the standard model. To resolve this conflict, we measured the efficiency of spatial frequency and orientation of filtered noise. To learn what information the visual system uses when detecting filtered noise, we applied a technique that can determine the information used to detect and discriminate filtered visual noise. In Chapter 2 the detection of spatial frequency filtered noise is not only efficient but remains so with stimulus uncertainty and extremely brief (10ms) stimulus duration. When the spatial frequency channel used was measured, we found a fixed bandwidth channel as the spatial frequency of the pattern was increased. To test the standard model, we implemented simulations of the standard model and contrary to the interpretation, the standard model could predict detection of spatial frequency filtered noise. Chapter 3 used spatial frequency filtered noise to relate the detection and discrimination of filtered visual noise. A simple rule relates what information observers use to detect and discriminate spatial frequency filtered noise. Chapter 4 extends the work of Chapter 2 to orientation information and found that orientation filtered noise is detected efficiently. We again measured what information observers used and found that, unlike SF filtered noise, observers use orientation in a flexible or adjustable manner. / Thesis / Doctor of Philosophy (PhD)

With Canada's increasing population, natural disasters such as flooding events will have an increasing impact on human populations. The severity of these events requires that decision makers have a clear understanding of the flood risks that communities face in order to plan for and mitigate flood risks. One key component to understanding flood risk is flood exposure, an element of which is the presence of structures (e.g., residences, businesses, and other buildings) in an area that could be damaged by flooding. Presently, several resources exist at both the national and global level that can be used to estimate the spatial distribution of structures. These resources are typically generated at global scales and do not account for regional or local data or processes that could enhance the accuracy and precision of exposure estimation in sparsely populated areas. The present study investigates the feasibility of creating a region-specific dwelling distribution model that helps improve estimation of residential structures in rural areas. Herein, we describe a rural dwelling distribution model for the province of Alberta that can be used to assist in the estimation of structural exposure to flood risk. The model is based on a random forest classification algorithm and several publicly available datasets associated with dwelling and population density. The model was validated using visually referenced data collected from earth imagery. The resulting dwelling layer was then evaluated in its ability to spatially disaggregate census dwelling counts, as well as predict dwelling exposure in several scenarios. This method appears to be a useful alternative to globally scaled models, or using the census alone, particularly for rural areas of Canada. / Thesis / Master of Science (MSc)

An increasing volume of empirical literature demonstrates the possibility of spatial autocorrelation in land value models. A number of objections regarding the methodology followed in those empirical studies have been raised. This thesis examines three propositions. The first proposition states that there is spatial dependence in the land value model in Roanoke County. The second proposition is that mechanical construction of neighborhood effects, or grouping nearby land parcels into neighborhoods, is not always the best way to capture spatial effects. Finally, the third and most important proposition states that by implementing a comprehensive set of individual and joint misspecification tests, one can better identify misspecification error sources and establish a more statistically sound and reliable model than models based on existing spatial econometric practices. The findings of this dissertation basically confirm the validity of those three propositions. In addition, we conclude that based on their development status prices of land parcels in Roanoke County may follow different stochastic processes. Changes in the values of hedonic variables have different implications for different groups of land parcels. / Ph. D.

This dissertation comprises three essays on theoretical and empirical issues in industrial organization. Chapter 1 outlines the issues explored in the subsequent chapters and briefly describes their conclusions. Chapter 2 explores how product differentiation impacts the incentive compatibility condition for firms to sustain implicit collusion in games of repeated interaction where, in contrast to previous studies, I focus on a market which is simultaneously vertically and horizontally differentiated. To achieve this objective, vertical differentiation is incorporated into an otherwise standard Hotelling framework. The ensuing mixed model of differentiation shows how the interrelationships between both forms of differentiation impact the incentives to collude, and is more general since it replicates previous findings throughout the literature. In Chapter 3, a multiproduct oligopoly model admitting product differentiation and a discrete choice demand model are proposed and estimated to determine if patterns of anti-competitiveness exist across distinct segments of the European car market. This chapter focuses on the evolution of price competition at a finer level than has been studied with a view to empirically challenge the notion that the European car market is wholly anti-competitive. Empirical results show that firm conduct varies due to the intensity of within-segment competition among rival firms. There is evidence of softer competition in the larger, mid- to full-sized segments and more aggressive competition in the smaller, entry-level subcompact segment. Chapter 4 represents a formal extension of the analysis in Chapter 3. In this chapter I examine the competitive structure of the U.S. automobile market using proprietary data comprising actual dealer-level transaction prices of several models of cars and light trucks sold in the domestic U.S. market between 2004 and 2007. The chapter is the first such study to employ consumer end-prices for automobiles in a structural New Empirical Industrial Organization (NEIO) framework. Empirical results reveal that there is more aggressive pricing in the light truck segments comprising minivans/SUVs and pickups, Bertrand pricing in the smaller, entry-level car segments, and softer competition in the full-size car segment. There is also a strong preference for domestically produced light trucks although consumers generally prefer to drive fuel efficient vehicles. / Ph. D.

Visual learning experiences are becoming increasingly prevalent in education as symbols, imagery and simulations replace traditional text-based materials. Although the utilization of images for instructional purposes is not a new occurrence, most images used in instruction have been two-dimensional representations, giving learners little experience working with three-dimensional images. Little research has been done to explain the effects of two-dimensional and three-dimensional stimuli on the learning process. This study examined the effects of two-dimensional and three-dimensional stimuli on spatial representation in drawings. Through the use of stereopsis, a scene was projected as both a two-dimensional image and as a three-dimensional image. Students wore polarizing glasses to enable them to perceive the superimposed images as a three-dimensional scene; whereas a single slide was projected when the image was to be perceived as a two-dimensional scene. Four test groups were established from eighth grade students who elected to take art. Participants in Group A were ask to draw the scene from the two-dimensional stimulus and, a week later, from the three-dimensional stimulus. Group B was asked to draw the scene from the three-dimensional stimulus and, a week later, from the two-dimensional stimulus. Group C drew only from the two-dimensional stimulus while Group D drew only from the three-dimensional stimulus. In all groups, participants were asked to draw the scene as realistically as possible using a graphite pencil. The completed drawings were evaluated for evidence of spatial cues and the students' perception and response to spatial information. / Ph. D.

Spatial referencing denotes the act of referring to a location or an object in space. Since it is often essential in different collaborative activities, good support for spatial referencing could lead to exceptional collaborative experience and performance. Augmented Reality (AR) aims to enhance daily activities and tasks in the real world, including various collaborations and social interactions. Good support for accurate and rapid spatial referencing in collaborative AR often requires detailed environment 3D information, which can be critical for the system to acquire as constrained by current technology. This dissertation seeks to address the issues related to spatial referencing in collaborative AR through 3D user interface design and different experiments. Specifically, we start with investigating the impact of poor spatial referencing on close-range, co-located AR collaborations. Next, we propose and evaluate different pointing ray techniques for object reference at a distance without knowledge from the physical environment. We further introduce marking techniques aiming to accurately acquire the position of an arbitrary point in 3D space that can be used for spatial referencing. Last, we provide a systematic assessment of an AR collaborative application that supports efficient spatial referencing in remote learning to demonstrate its benefit. Overall, the dissertation provides empirical evidence of spatial referencing challenges and benefits to collaborative AR and solutions to support adequate spatial referencing when model information from the environment is missing. / Doctor of Philosophy / People often exchange spatial information about objects when they work together. Example phrases include: ``put that there'', or ``pick the third object from left''. On the other hand, Augmented Reality (AR) is the technology that displays 3D information into the real world to enhance or augment reality. Scientists and technology practitioners think that AR can help people collaborate in a better way. The AR system needs to have a good understanding of the physical environment to support exchanging spatial information in the first place. However, limited by current technology, acquiring spatial information from the real world is not always possible or reliable. In this dissertation, we first illustrate the severity of insufficient environmental knowledge when collaborators sit next to each other in AR. Then we present pointing ray techniques to help AR collaborators refer to distant objects without knowing where those objects are. We further explore different marking techniques that can help the AR system calculate the position of a point in space without scanning the area. Last, we provide an AR application that supports efficient spatial information communication in remote discussion around physical objects.

Land subsidence is the gradual settling or sudden sinking of the earth's surface. According to the United States Geological Survey more than 80% of identified subsidence in the United States is a result of groundwater removal. Due to the hydrologic structure and reliance on the Potomac Aquifer, the Southern Chesapeake Bay region of Virginia has suffered from land subsidence since the 1940s. In coastal regions, land subsidence can increase the risk of flooding. This paper presents a mathematical simulation that predicts land subsidence from groundwater pumping. This simulation is used to see how the location of groundwater pumping, as well as the amount of amount of groundwater pumped would differ from two different groundwater pumping policies. The first policy is aimed at limiting land subsidence in the region, while the second policy aims at limiting the damages from land subsidence. These two policies are used to show that a spatially heterogeneous groundwater pumping policy is necessary to minimize the damages from groundwater pumping when land subsidence is present. / Master of Science