reprint available: nonlinear scale-space filtering

[Issac Wong]

FTP-host: archive.cis.ohio-state.edu
FTP-file: pub/neuroprose/wong.scale_space.ps.Z

The file wong.scale_space.ps.Z is now available for
copying from the Neuroprose repository:

A Nonlinear Scale-Space Filter by Physical Computation
Yiu-fai Wong
Institute for Scientific Computing Research, L-426
Lawrence Livermore National Laboratory
Livermore, CA 94551
E-mail: wong at redhook.llnl.gov

Abstract---
Using maximum entropy principle and statistical mechanics, we derive and
demonstrate a nonlinear scale-space filter. For each datum in a signal, a
neighborhood of weighted data is used for scale-space clustering.
The cluster center becomes the filter output. The filter is governed by a
single scale parameter which dictates the spatial extent of nearby data used
for clustering. This, together with the local characteristic of the signal,
determine the scale parameter in the output space, which dictates the
influences of these data on the output. This filter is thus completely
unsupervised and data-driven. It provides a mechanism for a) removing noise; b)
preserving edges and c) improved smoothing of nonimpulsive noise.
This filter presents a new mechanism for detecting
discontinuities differing from techniques based on local gradients
and line processes. We demonstrate the filter using real images.
This work shows that scale-space filtering, nonlinear
filtering and scale-space clustering are closely related and provides a
framework within which further image processing, image coding and computer
vision problems can be investigated.

This work has been presented at IEEE Conf. Computer Vision and Pattern
Recognition and IEEE Workshop on Neural Networks for Signal Processing, 1993.

--Isaac Wong from Lawrence Livermore Lab