No description available.

No description available.

No description available.

No description available.

This thesis presents some substantial theoretical analyses and optimal treatments of Kohonen's self-organising map (SOM) algorithm, and explores the practical application potential of the algorithm for vector quantisation, pattern classification, and image processing. It consists of two major parts. In the first part, the SOM algorithm is investigated and analysed from a statistical viewpoint. The proof of its universal convergence for any dimensionality is obtained using a novel and extended form of the Central Limit Theorem. Its feature space is shown to be an approximate multivariate Gaussian process, which will eventually converge and form a mapping, which minimises the mean-square distortion between the feature and input spaces. The diminishing effect of the initial states and implicit effects of the learning rate and neighbourhood function on its convergence and ordering are analysed and discussed. Distinct and meaningful definitions, and associated measures, of its ordering are presented in relation to map's fault-tolerance. The SOM algorithm is further enhanced by incorporating a proposed constraint, or Bayesian modification, in order to achieve optimal vector quantisation or pattern classification. The second part of this thesis addresses the task of unsupervised texture-image segmentation by means of SOM networks and model-based descriptions. A brief review of texture analysis in terms of definitions, perceptions, and approaches is given. Markov random field model-based approaches are discussed in detail. Arising from this a hierarchical self-organised segmentation structure, which consists of a local MRF parameter estimator, a SOM network, and a simple voting layer, is proposed and is shown, by theoretical analysis and practical experiment, to achieve a maximum likelihood or maximum a posteriori segmentation. A fast, simple, but efficient boundary relaxation algorithm is proposed as a post-processor to further refine the resulting segmentation. The class number validation problem in a fully unsupervised segmentation is approached by a classical, simple, and on-line minimum mean-square-error method. Experimental results indicate that this method is very efficient for texture segmentation problems. The thesis concludes with some suggestions for further work on SOM neural networks.

This thesis describes experimental work done in art education with a small group of children that met each Saturday at Dallas Services for the Blind over a two year period. The research used textured collage art projects as the medium of artistic expression for the children.

La synthèse de textures est une technique qui génère une texture automatiquement grâce à un algorithme. Cette technique permet de réduire le temps de création des textures et le coût mémoire étant donné que seuls les algorithmes et leurs paramètres ont besoin d'être stockés. Cependant, des difficultés sont souvent rencontrées lors de l'utilisation des textures synthétisées. D'abord, les paramètres de ces textures sont difficiles à manipuler. Ensuite, l'algorithme de synthèse génère souvent les textures sous forme de tableaux de pixels nécessitant beaucoup de mémoire. Pour aborder ces difficultés, nous proposons les approches suivantes : pour améliorer la visualisation de l'espace des textures synthétisées, nous proposons de construire un résumé de cet espace: une seule image statique qui résume, dans un espace limité de pixels, les apparences produites par un synthétiseur donné. De plus, pour améliorer la sélection de paramètres, nous augmentons les curseurs qui contrôlent les paramètres avec des bandes visuelles révélant les changements qui se produisent quand l'utilisateur manipule les curseurs. Pour permettre à l'utilisateur d'interagir de manière interactive avec les résumés visuels, nous nous reposons sur un algorithme de synthèse par patch permettant de générer les textures de façon rapide grâce à une implémentation parallèle sur le processeur graphique. Au lieu de générer le résultat de ce synthétiseur sous forme d'un tableau de pixels, nous représentons le résultat dans une structure compacte et nous utilisons une méthode rapide permettant de lire des pixels directement à partir de cette structure / Texture synthesis is a technique that algorithmically generates textures at rendering time. The automatic synthesis reduces authoring time and memory requirements since only the algorithm and its parameters need to be stored or transferred. However, two difficulties often arise when using texture synthesis: First, the visualization and parameters selection of synthesized textures are difficult. Second, most synthesizers generate textures in a bitmap format leading to high memory usage. To address these difficulties we propose the following approaches: First, to improve the visualization of synthesized textures we propose the idea of a procedural texture preview: A single static image summarizing in a limited pixel space the appearances produced by a given synthesizer. The main challenge is to ensure that most appearances are visible, are allotted a similar pixel area, and are ordered in a smooth manner throughout the preview. Furthermore, to improve parameters selection we augment sliders controlling parameters with visual previews revealing the changes that will be introduced upon manipulation. Second, to allow user interactions with these visual previews we rely on a fast patch-based synthesizer. This synthesizer achieves a high degree of parallelism and is implemented entirely on the GPU. Finally, rather than generating the output of the synthesizer as a bitmap texture we encode the result in a compact representation and allow to decoding texels from this representation during rendering

by Ho Kam-ho. / Chinese title: / Bibliography: leaves 133-134 / Thesis (M.Phil.) -- Chinese University of Hong Kong, 1983

This work started with a shape and a technique. Finding a method in developing volumes with the material in focus has been a theme throughout the project. The boundary between material and form has been investigated, just as the transition from sketch into ready garment. The result is a colourful and expressive collection of garments for women. The composition of the collection has been put together much as when making a collage, adding one interesting piece to another. The final collection ranges from one-off crafted pieces to more commercial or reproducible garments. / Program: Modedesignutbildningen

Thesis (M.S.)--Oregon State University, 2008. / Printout. Includes bibliographical references (leaves 100-106). Also available on the World Wide Web.