Iterative Noise Elimination Preliminary Report.
Abstract
A common feature of the empirical data in applied mathematics is a low signal to noise ratio. Define loosely as the ratio of the magnitude of essential signals to the magnitude of the worst noise, it often runs between 1/10 and 1/15, as was the case in the early work on computer tomography with digitized medical x-ray films which was the origin of this project. In such situations, straight-forward linear smoothing procedures tend to broaden the noise peaks without reducing the magnitude sufficiently. A common redeeming feature of the noise is that the largest peaks are quite sparse, lesser peaks are more frequent but still sparse, etc. This suggests a stepwise smooting in which only the worst peaks are removed during the early steps when there is no very good way to remove them. The purpose of the present report is to describe the procedure and to show the results of a series of tests with data from a computed tomography x-ray scan of a defective batery.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1983
- Accession Number
- ADA127756
Entities
People
- K. T. Smith
Organizations
- University of Wisconsin–Madison