Nonlinear Filtering Using Linear Combinations of Order Statistics.
Abstract
A nonlinear filter is considered whose output is given by a linear combination of the order statistics of the input samples. Assuming a constant signal in white noise, the coefficients are chosen to minimize the output MSE for several noise distributions. This new general filter is superior to both the median filter and the averaging filter, since they are just special cases. A class of heavy-tailed noise distributions is then considered, and optimal coefficients are found for various parameter values to determine the dependence of the filter coefficients on the heavy-tailedness (impulsivity) of the additive noise. Finally, some examples of designed filters are given operating on some simple inputs, and an examination of the root sets of the general filter is given. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1982
- Accession Number
- ADA124459
Entities
People
- Alan Conrad Bovik
Organizations
- University of Illinois Urbana–Champaign