Correlation on Noisy Images
Abstract
A computer program can simulate star-shaped images and a correlation between two different images on 128 by 128 matrices. The Fourier transform, in a time-saving algorithm, is used to carry out the correlation. Different factors in the image (contrast, noise, size) are varied and the effects are observed. The results showed good correlation except when the contrast was low and the noise was high. Using a filtering function in the correlation process produced good results in improving poor correlation patterns.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1980
- Accession Number
- ADA088659
Entities
People
- Eugene A. Margerum
- James A. Shine
Organizations
- Geospatial Research Laboratory