Retina located in the back of the eye is not only a vital part of human sight, but also contains valuable information that can be used in biometric security applications, or for the diagnosis of certain diseases. In order to analyze this information from retinal images, its features of blood vessels, microaneurysms and the optic disc require extraction and detection respectively. We propose a method to extract vessels called MF-FDOG. MF-FDOG consists of using two filters, Matched Filter (MF) and the first-order derivative of Gaussian (FDOG). The vessel map is extracted by applying a threshold to the response of MF, which is adaptively adjusted by the mean response of FDOG. This method allows us to better distinguish vessel objects from non-vessel objects. Microaneurysm (MA) detection is accomplished with two proposed algorithms, Multi-scale Correlation Filtering (MSCF) and Dictionary Learning (DL) with Sparse Representation Classifier (SRC). MSCF is hierarchical in nature, consisting of two levels: coarse level microaneurysm candidate detection and fine level true microaneurysm detection. In the first level, all possible microaneurysm candidates are found while the second level extracts features from each candidate and compares them to a discrimination table for decision (MA or non-MA). In Dictionary Learning with Sparse Representation Classifier, MA and non-MA objects are extracted from images and used to learn two dictionaries, MA and non-MA. Sparse Representation Classifier is then applied to each MA candidate object detected beforehand, using the two dictionaries to determine class membership. The detection result is further improved by adding a class discrimination term into the Dictionary Learning model. This approach is known as Centralized Dictionary Learning (CDL) with Sparse Representation Classifier. The optic disc (OD) is an important anatomical feature in retinal images, and its detection is vital for developing automated screening programs. Currently, there is no algorithm designed to automatically detect the OD in fundus images captured from Asians, which are larger and have thicker vessels compared to Caucasians. We propose such a method to complement current algorithms using two steps: OD vessel candidate detection and OD vessel candidate matching. The proposed extraction/detection approaches are tested in medical applications, specifically the case study of detecting diabetic retinopathy (DR). DR is a complication of diabetes that damages the retina and can lead to blindness. There are four stages of DR and is a leading cause of sight loss in industrialized nations. Using MF-FDOG, blood vessels were extracted from DR images, while DR images fed into MSCF and Dictionary and Centralized Dictionary Learning with Sparse Representation Classifier produced good microaneurysm detection results. Using a new database consisting of only Asian DR patients, we successfully tested our OD detection method. As part of future work we intend to improve existing methods such as enhancing low contrast microaneurysms and better scale selection. In additional, we will extract other features from the retina, develop a generalized OD detection method, apply Dictionary Learning with Sparse Representation Classifier to vessel extraction, and use the new image database to carry out more experiments in medical applications.

This thesis is divided into two parts, each of which examines aspects of bloodstain analysis where gravity is the main force applied to blood. Part I is a preliminary study on the dynamics of blood flow on various inclined surfaces and examines the use of blood analogs for easy test replication. The flow of uncoagulated human blood at different volumes and temperatures was examined on wood at a set angle of 1.5°, and on glass at varying incline angles. Glycerol solutions of 59% and 42% were used to represent blood at 23°C and 37°C respectively. Glycerol flow trials of similar volumes were conducted on wood, PVC and glass. Fluid flow plots of distance versus time exhibited double exponential curve behaviour, although a power-law relationship derived by H. E. Huppert's (1982) flow expression was obtained for blood flowing on inclined wood. Blood flow exhibited several observable characteristics; a decreasing width of the leading edge over time, and streaking and component separation of the leading region at very low speeds. On a glass surface, the width of the initial flow region decreased and initial speed increased with increasing angles. The glycerol analogs used in this study did not represent their blood counterparts well due to differences in physical properties of the fluids. Part II of this study focuses on the forensic value of passive bloodstains on three fabrics; 100% cotton drill, 65/35 polyester cotton, and 100% Shantung silk. 26 µL drops of 37°C human blood were deposited onto the three fabrics and paper from a height of 14 cm at various impact angles. The stains were photographed and analysed qualitatively and quantitatively using computational methods. 100% cotton drill, 65/35 polyester cotton and ironed 100% Shantung silk provided useful forensic values such as direction of travel and angle of impact. Overall, this study has provided useful preliminary data for further research work.

The role of language has been identified as crucial in the cognitive development of young children, and has been observed on different time-scales. In particular, the real-time verbal mediation of behaviour has been studied in the context of private speech use and self-regulation, pioneered by Vygotsky and continued by others who followed this line of research. However previous studies have mainly attempted to find correlations between speech and self-regulatory behaviour, but have been unable to capture the dynamic and real-time temporal interactions between these phenomena. Hence, without being able to carry out a contextual analysis of the actual instances of temporal co-occurrence between speech and behaviour, correlational analysis is limited in determining the various kinds of verbal mediation that children spontaneously employ as strategies during problem-solving and while exercising self-regulation. The current study proposes ‘temporal pattern analysis’ as an effective method of extracting significantly recurring patterns of task-relevant speech and goal-directed behaviour, as they repeatedly occur in a stream of naturalistic behaviour which may also contain other temporally random events. These recurring temporal patterns are then contextually analysed, considering the pragmatic content of the speech involved and the goal-directedness of the behaviour towards a specific goal of the episode. Goal-directed episodes of behaviour in eight typically-developing preschool children were video-recorded during their self-initiated activities in the classroom as well as during a problem-solving task held in a laboratory setting. The proposed method of temporal and contextual analysis was used to examine the role of both private as well as social speech in the verbal mediation of self-regulatory behaviour during goal-attainment. A Contextual Model of Verbal Mediation was proposed in the study to account for the diverse functions that both social and private speech perform during verbal mediation of one’s own and others’ behaviour in a goal-directed setting, depending on the specific social and task-related context. A dynamic framework of assessment of performance was developed in the study, to account for both successful attempts at self-regulation as well as failures of self-regulation. The study also attempted to determine any consistent group differences in the styles of verbal mediation employed by the children, across the classroom and the laboratory settings.

Denna studie presenterar en artefakt som med hjälp av Microsoft Kinect, samlar in och beräknar biomekanisk data från golfsvingar för att identifiera nyckelpositioner på ett automatiserat sätt. Den övergripande metoden som används är design-science research. Kinect sensorn är egentligen gjord för tv-spel, men kan även användas allmänt för att fånga och samla in kvantitativ biomekanisk data. Sensorn är inte specifikt utformad för golfsvingar, och saknar dessutom förmågan att spåra externa objekt som golfklubbor. Dessa problem var grunden för denna studies övergripande syfte, nämligen att utveckla en mjukvara med förmågan att identifiera de fem viktigaste nyckelpositionerna i golfsvingen. Nyckelpositionerna definieras utifrån mätbara egenskaper vilket för nyckelpositionen impact, krävde att man utnyttjade sensorns förmåga att spela in ljud. I empirin som genomfördes på en driving range samlades data från sammanlagt 20 svingar in. Varje identifierad nyckelposition analyserades på ett kvalitativt sätt utifrån ett antal sammanställda kriterier. Kinectsensorn hade problem att identifiera vissa positioner, men sammantaget bedömdes 87 % av de insamlade nyckelpositionerna som korrekt identifierade. För nyckelpositionen impact, där insamlad ljuddata användes för identifiering, bedömdes 85 % av de insamlade nyckelpositionerna som korrekt identifierade. Studien begränsar sig till utvalda nyckelpositioner men visar potential för automatiserad insamling av kvalitativ golfsvingdata, och ger uppenbara möjligheter för vidare forskning. / This study presents an artifact that is using Microsoft Kinect for motion capturing of golf swings, in order to identify key positions in an automatic fashion. The main method used is design science research. The Kinect sensor, which is developed for the Xbox video game console, can be used for general motion capture. It is, however, not tailored for golf swings and it also lacks the ability to track external objects such as the golf club. These problems were the main motivation for the purpose of this study, i.e. to develop an application to identify the five most important positions in the golf swing. The key positions were defined based on measurable traits, making it necessary to use the audio recording ability of the sensor for the impact position. The empirical investigation was performed at a driving range, and data from a total of 20 golf swings were gathered. In the next step, every key position was analyzed in a quantitative manner based on a number of criteria. The results show that the Kinect sensor may have some troubles recognizing certain positions, but still 87 % of the key positions captured were considered to be successfully identified. Specifically, 85 % of all impact positions were successfully identified. The study was limited to the chosen key positions, but shows good potential for automatic capturing of quantitative golf swing data, thus suggesting several possible directions for future research.

No description available.

The aerospace industry endeavours to improve modern aircraft capabilities in efficiency, endurance, and comfort. One means of achieving these goals is through new enhancements in aerodynamics. Increased wing aspect ratio is an example of further improving efficiency. However, this comes with new challenges including possibly adverse aero-elastic and aero-servo-elastic (ASE) phenomena. New computational methods and tools are emerging and there is a need for experimental data for validation. University of Victoria’s Centre for Aerospace Research (UVic CfAR) set out to design a 20kg ASE demonstrator using a remotely piloted aircraft (RPA). This aircraft was designed with the intent of exploring coupling between aero-elastic modes including coupling between the short period aerodynamic mode and the first out-of-plane elastic mode of the wing. This thesis discuses the implementation of instrumentation designed and integrated into the ASE RPA demonstrator to monitor the deformation of the elastic wing in-flight. A strain based measurement technique was selected for integration into the ASE aircraft. This choice was made for several reasons including its reliability regardless of outdoor lighting, relatively lightweight processing requirements for real time applications, and suitable sampling bandwidth. To compute the wing deformation from strain, a method, sometimes referred to as strain pattern analysis (SPA), utilizing linear combinations of reference modal shapes fit against the measured strain, was used. Although this method is not new, to the author’s knowledge, it is the first practical application to a reduced scale RPA demonstrator. The deformation measurement system was validated against a series of distributed static load tests on the ground. Distributed load cases along the wing demonstrated good out-of-plane measurement performance. A case where only load is applied near the root of the wing resulted in the largest error in part as the mode shapes generated are less suited to approximate the resulting shape. In general errors in out-of-plane displacement at the end of the flexible wing portion can be expected to be less than 5%. The displacement at the tip of the wing can be as great as 11% for the left wing whereas the right wing is 4.7%. This suggest an asymmetry between the left and right wings requiring specifically tuned FE models for each to achieve best results. Twist angles presented in tests were relatively small for accurate comparison against the reference measurement, which was relatively noisy. Generally, the deformation measurement by SPA technique followed the same twist behaviours as the reference. A twist case, unlikely to be seen in flight, provided some insight into twist measurement robustness. The work presented is merely a small step forward with many opportunities for further research. There is room for improvement of the FE model used to generate the mode shapes in the strain pattern analysis. Initial efforts focused on the flexible spar portion of the wing. With more work improvements could be achieved for the estimation of the rigid wing. Additionally, there was some asymmetry between each wing semi-span, and with some focus on the left wing its results could be improved to at least match that of the right wing. A real-time implementation was not completed and would be particularly interesting for use as feedback for flight control. Study of load alleviation techniques may benefit. Another topic of study is the combination of this method with other measurements, such as accelerometers, to provide improved performance state estimation through sensor fusion. / Graduate

Learning outcomes for students whose education takes place in the physical field are, among others, knowledge and skills to implement the occupational therapy process. In this process patients’ problems are assessed and recorded, treatment planned, implemented, continuously evaluated, and the progress of each such patient professionally recorded. Since this is a process which requires distinct clinical reasoning skills on the part of the student various factors can influence the development of such skills during their training. Many authors are of the opinion that it is the interpersonal communication between supervisor and student which underpins successful fieldwork education. In this study the purpose therefore was to investigate how the interpersonal communication factors in the supervisory relationship play a role in enhancing occupational therapy students’ clinical reasoning during physical fieldwork education. At the outset a partially mixed, sequential dominant, status-qualitative design was employed. An inter-subjective or interactional epistemological position was adopted in order to generate data from the participants’ subjective experiences, and an interpretive approach was used to understand how occupational therapy students and supervisors perceive the supervisory relationship during the formers’ learning of their clinical reasoning skills. Data was generated from four sources. First of all, from focus groups conducted separately with students and their supervisors on completion of the fieldwork block; secondly from semi-structured one-on-one interviews held with students as well as supervisors on completion of the formers’ fieldwork block; thirdly from students’ Work Habits Reports, and finally by recording the practical exam grades students obtained in the physical field. To analyse the data both qualitative and quantitative research methods were employed. Information obtained from the focus groups and one-on-one interviews were audio-taped and transcribed. After this process, transcribed data was coded and analysed following both a bottom-up and top down approach. The former was carried out by an independent coder and the latter by the researcher herself to determine which interpersonal communication themes and patterns might emerge from the collected data. A clinical psychologist using the Interpersonal Pattern Analysis, a diagnostic instrument, analysed the audio tapes of 14 supervisors who participated in the focus groups and one-on-one interviews. The themes which emerged from the thematic-content analysis and the Interpersonal Pattern Analysis were compared with the grades students obtained for their clinical reasoning skills in the final practical exam in the physical field. The findings of this study indicated that supervisors of students who received high grades solved problems effectively, were predominately linear in their approach, showed only limited empathy, were rigid in their expectations and gave only limited confirmation. In line with these findings supervisors of students who received lower grades were also effective in terms of problem solving skills and also gave limited confirmation, but were circular in their approach, showed partial empathy and were flexible. Finally in respect of the interpersonal approach to human behaviour there is no one role or pattern of interaction that is more effective in all contexts. A style or a pattern that may be highly effective in one kind of relationship may be ineffective in another. What is emerging here though is that a style which is characterised by flexibility and empathy may not necessarily be an effective teaching style, whereas a style characterised by a linear approach and limited empathy did indeed prove to be significantly more effective. / Thesis (PhD)--University of Pretoria, 2011. / Occupational Therapy / unrestricted

No description available.

Agricultural soil properties exhibit variation over field plot scales that can ultimately effect the yield. This study performs multiple spatial pattern analyses in order to design spatially dependent regression models to better understand the interaction between these soil properties. The Cation Exchange Capacity (CEC) and Calcium-Magnesium Ratio (CaMgR) are analyzed with respect to Calcium, Magnesium, and soil moisture values. The CEC and CaMgR are then used to determine impact on the yield values present for the field. Results of this study show a significant measure of model parsimony (0.979) for the Geographically Weighted Regression (GWR) model of the CEC with free Ca, Mg, and soil moisture as explanatory variables. The model for CaMgR using the same explanatory variables has a much lower measure of model fit. The yield model using the CEC and CaMgR as explanatory variables is also low, which is representative of the underlying processes also impacting yield.

Methods are needed to predict the spatial distribution pattern(s) of forest harvests to improve large-scale timber inventories through the incorporation of spatial distribution information. In this study, land cover type change detection methods were carried out using information from 1972 thru 2005 east-central Mississippi Landsat TM data to detect the spatial distribution of clearcut forest harvests. A random pattern was observed for clearcut harvests and this and other statistical data gathered in the study will be integrated into an existing forest inventory and transportation network database. The resulting harvested and forested area predictions generated by the database will assist with determining the sustainability and availability of forest resources for existing and future forest product mills and the creation of transportation networks needed to supply raw materials and distribute end products. Other implications include assessment of the quality and spatial components of wildlife habitat and a tool for forest certification programs.