PDA

View Full Version : Re: Digitizing marker locations in 3D cinematography (J)



Jaekun Shim
04-08-2002, 03:59 AM
Dear Jonathan Merritt,

I am Jaekun Shim, a graduate student working for Ball State University
Biomechanics Lab. I just hope that my answers will help you developing
your digitizing software.

|However, I have also come across automatic
|digitizing methods which are advertised as part of commercial gait
|analysis systems. So far, I have not been able to find a good
reference
|for how these automatic systems work.

Ans>
Even for the automatic digitizing, you generally need to digitize the
markers on the first frame. After you digitize those, a program
registers arrays corresponding to the markers. Once you know where the
markers are on (u, v) coordinate system (pixel units on your screen),
then you will be able to use "SEARCH RANGE" (the program is looking for
the marker on the second frame in a certain diameter from the previous
marker) or "MARKER SIZE" (program recognize a blob as a marker only if
the marker is bigger than the user-defined marker size). I think it will
be a better algorithm if you can use "quadratic (or linear)
extrapolation" technique to predict the marker location.

You may want to check "Blob Analysis" on the web. There is tons of
information there about it. Blob analysis includes the determination of
geometrical properties such as area, perimeter, and orientation of each
blob as well as labeling and counting the blobs in the image.

|Intuition says that it should be possible to use some sort of
|thresholding or watershed algorithm to find marker locations, using the
|calculated centroid of the marker as "the location"

Ans>
Since we do not need to use a color image but a grayscale image, the
image for digitizing is usually "Binarized" using a user-defined
threshold. For example in bitmap image, after you choose a threshold,
the pixels on the image whose gray values are greater or equal to the
threshold are replaced with 255 (which is white on bitmap image) and the
other pixels are changed with 0 (which is black on bitmap image). I
believe that this process will make automatic digitizing much efficient
and easier.

You will be able to see the difference between two images before and
after the Binarization using a threshold by clicking on the followings;
Before: www.bsu.edu/web/jkshim/download/walking.bmp
After: www.bsu.edu/web/jkshim/download/binarizedwalking.bmp

Good luck!

J

_______________________________________

Jaekun Shim
Ball State University,
Biomechanics Lab, Human Performance Lab
Muncie, IN 47306, U.S.A.
765)285-5178
shimjaekun@hotmail.com jkshim@bsu.edu
www.bsu.edu/web/jkshim



|-----Original Message-----
|From: Biomechanics and Movement Science listserver [mailto:BIOMCH-
|L@NIC.SURFNET.NL] On Behalf Of Jonathan Merritt
|Sent: Saturday, April 06, 2002 11:17 PM
|To: BIOMCH-L@NIC.SURFNET.NL
|Subject: [BIOMCH-L] Digitizing marker locations in 3D cinematography.
|
|Hello Everyone,
|
|I am currently researching the basics of 3D cinematography systems that
|use retroreflective markers. My eventual aim is to write some software
|that can take scanned film images from a set of cameras and produce a
|set of 3D coordinates for the markers.
|
|Presently, I am looking for methods by which markers may be digitized
|automatically in two-dimensions from film frame images. I am aware
that
|one method by which markers are digitized is simply for an operator to
|view each frame of a motion-capture session and manually tag the marker
|locations with a cursor. However, I have also come across automatic
|digitizing methods which are advertised as part of commercial gait
|analysis systems. So far, I have not been able to find a good
reference
|for how these automatic systems work. I have searched the ISB software
|pages, the BIOMECH-L archives and the internet. However, I am slightly
|unsure about which field encompasses this question and what keywords I
|should be using.
|
|Intuition says that it should be possible to use some sort of
|thresholding or watershed algorithm to find marker locations, using the
|calculated centroid of the marker as "the location" (and what I have
|found so far hints at this). Does anyone know of any good papers on
|this subject, or what keywords I should be using in order to find them?
|
|Many thanks, :-)
|
|--
|Jonathan Merritt, PhD Student.
|Equine Clinical Research Unit,
|Faculty of Veterinary Science,
|The University of Melbourne.
|
|---------------------------------------------------------------
|To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
|For information and archives: http://isb.ri.ccf.org/biomch-l
|---------------------------------------------------------------

---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------