The Fuzzy Page

To See Clearly using Fuzzy Vision:
Applications of Fuzzy Methods to New and Old Problems of Computer Vision

This work involves segmentation of images using fuzzy models. Planned are methods for pixel, edge and region-based schemes, that use several fuzzy clustering models.

Downloadable Publications
Harvey A. Cohen, The Gerrymander Problem in Vector Quantization, Proc, Australian and New Zealand Conf. Intelligent information Systems, ANZIIS-96, Adelaide, Nov 18-20, 1996, pp 308-311. In a political gerrymander, some elected politicians represent electorates with huge populations, while others represent electorates of small population. In image analysis and in image engineering, there is a process that is very much akin to holding an election, it's called code book coding, and is also known as vector quantization. In this process, image blocks, typically 4x4, which are called vectors, are labelled with the like blocks (= code vectors) of a code book. In other words, each block in an image is represented by a particular block in the code book, When things go well, the code book vectors are pretty representative of the population, but not always. Someimes some code book vectors represent many members of the population of vectors, while some represent just a few r none at all; in other words, a gerymander applies.
This paper is devoted to explaining just what a gerrymander in VQ looks like, and, with a bit of insight from the more famous political gerrymanders, a way around gerrymanders is described.
Paper 96:10 at my Paper Download list.

J. You and H.A. Cohen, Textured Image Segmentation by Optimization of Fuzzy Compactness, Australian Pattern Recognition Society, Segment’96, Sydney Dec 13, 1996, pp 97-100. Paper 96:13 at my Paper Download list.

Link to Dr Harvey Cohen: Refereed Conference and Journal Papers 1989-1997

Link to Image Index