3D Photography: Vision Lab Informations
You will find in this page some advices for the lab.
Please, read this page carefully and regularly.
The project Report:
You can download a tar version of the report
template directory. The main file to load is report.html.
1- Grabbing images
On the linux machine, there is a program that acquires simple
sunraster files.
The syntax is:
acquire image.ras
It dumps the data in the file image.ras that can be read using xv
or readras.m (under matlab ).
The source code (in C) and the Makefile can be found on the
/home/ftp/pub/dt3155/ directory of the machine.
The executable is copied in /usr/local/bin/ where you can also
find a simple script file objac* (from Emil and Marcel) that
allows multiple image acquisition. From that example you can write
your own script very painlessly.
You can also get the tar version of the dt3155
directory online.
2- Remote access to the
machine
You can remotely access to the machine by doing ftp or
telnet to photo3d.caltech.edu (the IP address is 131.215.9.74).
3- Some useful matlab functions
There are available a couple of matlab programs on the photo3d
machine (in /home/ftp/pub/):
- click_points.m: Helps you select points on an
image at sub-pixel accuracy. Syntax: [x] = click_points(I,map),
where (I,map) is an image. Click (approximately) on the figure 1
on the sequence of points you want to select (with the mouse button
1), when done, press Return. The figure 2 shows the succession of
close up views of the clicked points, and waits for more
accurate clickings. The output is the matrix of selected image points.
- comp_distortion.m: The inverse function of the radial
distortion transformation. Syntax: [x] = comp_distortion(x2,k)
where x2 is the matrix of image points with the radial distortion
(after substracting out the camera center position, and dividing by
the focal lengths) and k is the radial distortion coefficient.
The output x is the matrix of the points without distortion.
There is now a Automatic Calibration Package available online!!!
Retrieve the compressed tar file AUTOCALIB.tar.gz.
You will find in the README file some instructions as how to use it
for doing the full camera and projector calibrations in 8 clicks total!!
4- Programs to generate Stripes on the Laptop
Thanks to Emil and Marcel, there are a couple of programs
on the laptop that generates stripes on the screen for the projector
(in the D partition in the directory D:\stripes\).
Take a look at them, starting with stripes.exe and checker.exe.
There is another program that generates directly gray code stripes
(Gcodestr.exe). Run it, and see what it does.
You should think of how those striped images are generated.
The Pascal sources files are available on the machine, but PLEASE
DO NOT MODIFY THEM!!! If you want, you can copy them to your account.
The Pascal compiler works. Therefore, you can write your own code for
generating anything on the projector screen. To compile a program,
type under DOS:
tpc myprog.pas
where myprog.pas is your Pascal source code.
(it turns out that Pascal is quite handy for doing fast graphical
stuffs)
Please DO NOT PUT A MESS IN THE LAPTOP, if you want, you can generate
your own directory on the D partition and put your stuff in it (name
it after your login name).
5- The SYNC issue for the projector
If the projector seems to project non uniformed stripes,
hit a few times the Sync button of the projector, and everything gets better.
If you change the SYNC of the projector and later on turn it off,
it loses the setup. A solution is to keep the projector on, and
only turn off the lamp, by pressing on the "lamp" button in front of
the projector.
Try not to keep the projector lamp on for long periods of time.
6- How do we increase the
projector depth of focus?
By lowering the aperture of the
projector, you significantly increase its depth of focus. Of
course, the trade off is brightness. Indeed, by lowering the aperture
we lower the brightness as well. In addition, if you make your
aperture too small, you introduce a bit of diffraction (a projectred
black blob).
I made up a couple of little devices to do the job. Please try them to convince
yourself of the effect(s). You can find them on or next to the projector.
7- Gain control of the
camera
The camera has some gain control. In other words, it
adjusts the overall gain on the image depending upon the average
illumination. Due to that effect, it is rather difficult to use the
absolute "white" and "black" images to locate the stripes on the
striped images. Therefore, I STRONGLY recommend to use the negative
of the stripes. This should not be so much affected by the camera
gain control. Think of why?
8- The equipment is expensive!!!
Take a good care of the equipment. All that is rather expensive.
Also, keep the COMBO code for yourself. We don't want the entire
campus to wander around the lab...
RETURN TO THE MAIN PAGE