These
tutorials were created for use in the short course "Computational
Neuroscience: Vision", held at Cold Spring Harbor Laboratories in Summer 1996. It is
provided "as is", without express or implied warranty, for
educational purposes only, and is not to be used, rewritten, or adapted as the
basis of a commercial software or hardware product. Please do not distribute this file without
prior permission of the instructors. Many hours were spent developing these
tutorials and we ask your compliance to assure proper attribution and
maintenance. If you would like to use
this tutorial in a course, please contact one of the following instructors:
Eero Simoncelli
<eero.simoncelli@nyu.edu>
Paul
Glimcher <glimcher@cns.nyu.edu>
EJ
Chichilnisky <ej@salk.edu>
Contains matlab
tutorials on image processing and vision.
Each tar file contains the tutorial plus the associated toolbox
directories. (For Windows use Winzip, and for Unix use tar -xvf Tutorial.tar).
Macintosh:
Put the entire Tutorials folder in the Matlab toolbox
folder. Then select "set path"
from the "File" menu and click on "set auto path". Matlab will then be
able to find the various mfiles in the TutorialsToolboxes folder.
Most of the tutorials are run by cutting and pasting Matlab
code from the tutorial file into the Command window. Open one of the tutorials, e.g., LinSysTutorial.m. Read the commented text. Use the mouse to mark a code segment. Then drag that code over to the Command
window and Matlab will execute it.
UNIX:
Put all of the subdirectories (e.g., color, filter, graphics, etc.) in the Matlab path. Most of
the tutorials are run by cutting and pasting Matlab
code from the tutorial file into the Command window. Open one of the tutorials, e.g., LinSysTutorial.m, in your favorite text editor. Read the commented text. Copy and paste the code segments into the Matlab window or buffer.
svdTutorial: A review of some basic concepts in
linear algebra, concentrating on the singular value decomposition. Run by cutting and pasting from the tutorial
file into the Matlab Command window.
LinSysTutorial: One-dimensional discrete Fourier transform
and filtering. Run by cutting and
pasting from the tutorial file into the Matlab
Command window.
SamplingTutorial: subsampling,
upsampling, and aliasing of one- dimensional discrete
signals. Run by cutting and pasting from
the tutorial file into the Matlab Command window.
ImageTutorial: Two-dimensional discrete Fourier
transform, filtering, and sampling. It
is recommended that you first do LinSysTutorial and SamplingTutorial. Run by cutting and pasting from the
tutorial file into the Matlab Command window.
ColorTutorial: Spectral power distributions, surface
reflectance functions, human cone spectral sensitivities,lots of examples of color calculations. Run by cutting and pasting from the tutorial
file into the Matlab Command window.
MotionTutorial: This tutorial will take you through a
simple derivation of an algorithm for computing optical flow, the projection of
motion onto the image plane. The
algorithm is based on measurements of the gradient, but, as we will show, may
also be thought of as a spatio-temporal
"Energy" algorithm
(ala Adelson/Bergen). Run by cutting and pasting from the tutorial
file into the Matlab Command window.
MotionEnergy: Implementation of the spatio-temporal
energy model described by Adelson & Bergen in
'Spatiotemporal energy models for the perception of motion' (JOSA-A,2:284-299). Run by
cutting and pasting
from the tutorial file into the Matlab Command window.
PyramidTutorial: Gaussian and Laplacian
pyramids, projection and basis functions of pyramid transforms, aliasing in
pyramid transforms Haar, QMF and wavelet pyramids, steerable pyramid.
Run by cutting and pasting from the tutorial file into the Matlab Command window.
StereoTutorial: Random-dot stereograms,
several algorithms (some considered reasonable models of human stereopsis) for estimating stereo disparity. Run by cutting and pasting from the tutorial
file into the Matlab Command window.
Presently unavailable:
MaskingTutorial: This tutorial takes
you through an example of how to set up and run, analyize
and fit a psychophysics experiment using MATLAB. This example works through a "fake"
spatial pattern detection experiment.
Rather than getting actual subject's responses, it uses a model to
simulate a subject's responses. Run by
cutting and pasting
from the tutorial file into the Matlab Command
window.
MTmodel_menu: Mike Shadlen's model for simulating a motion discrimination
experiment using simulated responses of MT neurons. Run by typing "MTmodel_menu" at the Matlab
prompt.
MissingFund: This tutorial illustrates an
influential motion illusion introduced by Ted Adelson
and Jim Bergen. The stimulus illustrates the superiority of a frequency-domain
based analysis of motion over the more common-sense approach of tracking the
change in location of conspicuous image features over time. There is one line of this tutorial that uses
the UCSBtoolbox (that runs only on the Macintosh) for
lookup table animation, but the rest of the tutorial can be run on any
platform. Run by cutting and pasting
from the tutorial file into the Matlab Command
window.
HodgkinHuxley: Simulink
implementation of the Hodgkin-Huxley model of action potential generation in
the squid giant axon. Run by typing "HodgkinHuxley" at the Matlab
prompt. HH-summary.ps
is a postscript file with all the equations and parameter values for the model.