Abstract

The statistical range was substituted for the variance in local noise smoothing algorithms proposed by Tomita and Tsuji, and by Nagao and Matsuyama in order to reduce computer processing time. Images of low, medium, and high information content were corrupted by two levels of combined additive and multiplicative noise and were processed by the original and modified algorithms for both the uncorrelated and correlated noise cases. A subjective paired comparison was performed on the resulting images based on four evaluation criteria, and a normalization of the data between observers as well as hypothesis testing were carried out. Under the conditions of the test, it was found that comparable noise smoothing performance could be achieved only with the algorithm by Tomita and Tsuji, but that noted improvements occurred in the areas of preservation of subtle details, immunity to shape distortion and preservation of step edges.

Library of Congress Subject Headings

Image processing--Digital techniques

Publication Date

8-1-1991

Document Type

Thesis

Department, Program, or Center

Chester F. Carlson Center for Imaging Science (COS)

Advisor

Granger, Edward

Advisor/Committee Member

Schott, John

Advisor/Committee Member

Engeldrum, Peter

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA1632 .L36 1991

Campus

RIT – Main Campus

Download

Share

COinS