Nonlinear Image Magnification:
My thesis research on nonlinear magnification shows great
potential as a tool for the enhancement of images. Very subtle
magnification techniques are possible through the application of
magnification brushing, which allows arbitrary regions
of an image to be expanded or compressed. The figure on the left
below shows an image of the martian landscape from the Mars
Pathfinder mission. The image on the right illustrates how
nonlinear magnification can be used to enhance the visualization
of the rock referred to as "Barnacle Bill". (Original image
Copyright (c) California Institute of Technology, Pasadena, CA.
All rights reserved.)
More information about nonlinear magnification can be found on my visualization research page or from the nonlinear magnification home page.
Wavelets:
I am interested in the application of wavelets to images. As part of a
research project, I implemented a version of the Discrete
Periodic Wavelet transform in Mathematica, which includes routines
to visualize the actual transform (i.e. that show the re/construction
of a signal from successive applications of the wavelet filter). In
addition to this code (which implements both 1D and 2D transforms),
there is also a mathematica notebook file which shows examples of how
to use the transform. The graphic below shows the result of
reconstruction from an identity matrix with Daubechies' D4 wavelet
basis. This graphic was produced with the above mentioned mathematica
wavelet tools.
I have also implemented C++ versions of the Fast Wavelet Transform
for interactive scanning of large image databases.
Wavelet and Image Processing Sites:
The Wavelet Digest
Contents
Computer Vision Home Page
Computer Vision Home Page (text only)
Wavelets at Imager
Research Group on Computer Vision and Artificial Intelligence, IAM, University of Bern
