Author

Wei Zhang

Publication Date

1994

Document Type

Dissertation/Thesis

First Advisor

Bow, Sing-Tze, 1924-

Degree Name

M.S. (Master of Science)

Legacy Department

Department of Electrical Engineering

LCSH

Neural networks (Computer science); Image processing--Digital techniques

Abstract

Image processing techniques have been applied in various areas. More and more applications require real-time image processing. Execution speed is the primary difficulty experienced in realizing real-time image processing, since there is a huge number of pixels which require processing. Current computer technologies do not support the required processing speed. Parallel processing is a very effective way to increase image processing speed. A neural network, which is composed of a large number of processing elements (PE), possesses the resources for parallel operations. It is, therefore, very advantageous to implement the image processing with a neural network. The work presented in this thesis is an application of neural network theory to the image processing in a highly parallel operation mode. Detailed algorithms and methods are proposed and discussed in the thesis. Various neural network architectures have been designed and evaluated for image convolution and enhancement. Intel’s neural network training system (iNNTS) was used for experiments in this thesis. A detailed discussion of the Intel 80170 neural chip with 64 neurons is included. Utilizing this hardware system, several neural network architectures have been applied to process image pixels in parallel for the image convolution and image enhancement. Experimental results obtained show that neural network model proposed in this thesis for image processing in parallel is effective. This model can be extended to a largescale neural network and real-time image processing can eventually be realized.

Comments

Includes bibliographical references (pages [88]-90).

Extent

x, 111 pages

Language

eng

Publisher

Northern Illinois University

Rights Statement

In Copyright

Rights Statement 2

NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.

Media Type

Text

Share

COinS