Bayesian Interpolation and Deconvolution
Abstract
The deconvolution problem is addressed in stages beginning with the interpolation problem when little prior information is available and proceeding to the full deconvolution problem when a great deal of prior information is available. The results of these calculations indicate that good solutions to the deconvolution program are available even when limited prior information is available and that these results overlap those obtained when a great deal of prior information is available. The difference between them is that the use of uninformative priors causes large uncertainties in the estimated signal, while highly informative priors decreases the uncertainties in the signal.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1992
- Accession Number
- ADA256185
Entities
People
- G. L. Bretthorst
Organizations
- University of Washington