Thesis (M.S.)--Oregon State University, 2010. / Printout. Includes bibliographical references (leaves 43-46). Also available on the World Wide Web.

Thesis (Ph.D.) - University of Queensland, 2004. / Includes bibliography.

Thesis (Ph.D.) - University of Queensland, 2005. / Includes bibliography.

Thesis (M.S.)--Rochester Institute of Technology, 2007. / Typescript. Includes bibliographical references (leaves 81-83).

Event related potentials (ERPs) are obtained from noninvasive electroencephalograms (EEG) which measure neuronal activity of brain on the scalp. However, conventional ERPs derived by averaging of single EEG trials have strong latency variability and are smeared, resulting in blurred scalp topography, especially in late components of ERP. The smearing problem had been addressed by reconstructing ERPs after latency correction with a new EEG analysis method Residue Iteration Decompo¬sition (RIDE), which was demonstrated in a face priming paradigm to improve distinctness in scalp topography (Ouyang et al., 2011). This thesis aims to (1) extend the benefits of RIDE to neural source space by localizing the neural generators of ERPs, thereby developing an integrated RIDE framework for improvement in source localization and causal modeling of effective source networks, and (2) apply the framework to the face priming paradigm for famous faces, to explore the dynamics of face processing and priming. We localized sources through brain electrical source analysis for both conventional ERP and RIDE derived ERPs (RERPs). RERPs allowed localization of an additional motor execution source (Premotor Cortex, PMC), apart from 5 other common sources, of which 2 (Occipital Lobe, OL; Fusiform Gyrus, FG) were obtained from early activity (< 250 ms) and 3 (Mediotemporal lobe, MTL; Prefrontal Cortex, PFC; Anterior Temporal Lobe, ATL) from late activities (> 250 ms) of RERPs respectively. Priming effects, i.e., the difference between primed famous (PF) and unprimed famous (UF) face conditions in source waveforms (SWFs), were extended and enhanced in RERPs, especially for late sources. The priming effects revealed (1) the role of sources in each hemisphere that play in perception, memory and execution, (2) parallel processing of information in sources, (3) early processing in the right hemisphere, and (4) predominance of the right hemisphere in face recognition. Results confirmed SWFs of RERPs as better choice for the dynamic causal model (DCM). Two candidate DCM models, forward (F) and forward-backward (FB) were outlined on each hemisphere with SWFs from PF and UF conditions of RERP data. Priming has tendency to facilitate the FB model in the left hemisphere. On the other hand, independent of model preference, priming strengthened a bidirectional connection between FG and PFC in both hemispheres; this indicates a strong role of FG in structural representation and of PFCs in controlling decisions about face familiarity. Priming modulates the pathway FGMTLPFC differently in the two hemispheres, strengthening the involvement of MTL in the left hemisphere and weakening in the right hemisphere. This indicates proficiency of the left and right MTL in processing different aspects of facial information. Further, a backward connection ATLPFC in the left hemisphere was found to be functionally relevant for both conditions in speeding up response time in individual subjects, reinforcing the role of PFC in executive functioning and ATL in naming of famous faces. Thus, an integrated framework of source localization and DCM with RERPs allows a novel, comprehensive understanding of time resolved dynamics in face recognition and priming, thereby piloting prospects of its application to other experimental paradigms.

Masters of Science / The communication barriers between deaf and hearing society mean that interaction between these communities is kept to a minimum. The South African Sign Language research group, Integration of Signed and Verbal Communication: South African Sign Language Recognition and Animation (SASL), at the University of the Western Cape aims to create technologies to bridge the communication gap. In this thesis we address the subject of whole hand gesture recognition. We demonstrate a method to identify South African Sign Language classifiers using an eigenvector ap- proach. The classifiers researched within this thesis are based on those outlined by the Thibologa Sign Language Institute for SASL. Gesture recognition is achieved in real- time. Utilising a pre-processing method for image registration we are able to increase the recognition rates for the eigenvector approach. / South Africa

Dimensional inspection with Vision Systems requires a careful selection of image analysis methods in order to obtain accurate information about the geometry of the parts to be measured. The purpose of this project is to study, implement and compare different image evaluation methods and to show their strengths and weaknesses with respect to dimensional inspection. Emphasis is made on the inspection of circular features. The criteria of comparison for these methods are discussed. Using synthetically generated images, various analysis methods are compared and conclusions for their use are drawn. Results of the comparison show that the selection of a method has to be done with regard to the noise level of the measurement. Finally, a computationally fast calibration algorithm is studied and implemented .

No description available.

No description available.

The problem of identifying three dimensional objects from their two dimensional perspective projections is an important one in computer vision. A segmentation procedure is described here to extract features from a simulated image, then a matching procedure which finds the three dimensional objects in the picture is discussed. An image: a photograph taken by a camera with fixed focal length, is given, and a three dimensional wire frame object model is also given. Each object in the model is composed of planar arcs. The arc is bounded by connected line segments or conics. The input image is a simulated photographs of some objects of our model, the problem is to identify what these objects are. The two dimensional data structure we derived from image are equations of each planar arc. The data stored in the database are equations of the boundary pieces of the three dimensional object. Our aim is to find the match between the given three dimensional data structure and the two dimensional data structure. The method used here is a numerical analysis nonlinear optimization. / Ph. D.