This is the current GANTT graph.
Run of project is shown by a different color (gray). So, you can see if
the project is leading or delayed. (update 21-07-2005)
Results
In RED, encountered
problems.
In GREEN, solutions.
22-05-2005 / 25-07-2005 :
Summer camp of the laboratory.
16-07-2005 / 22-05-2005 :
Ending the redaction of the report for Polytech'Grenoble
Displaying complex 3D
objects. Draw an object in OpenGL is pretty long and we can't freeze in
binary all organs. Thus, we need to import organs separately.Installation of ARToolKit 2.70 (the
last one, with VRML2.0 support). Rewritting of previous program to fit
the new specifications.
Modeling an organ (lung). Data
didn't fit WRML2.0 format.Installation of Blender (3D modeler).
Here is a screenshot :
The camera was capturing 3 markers.
The WMRL Lung model is display in the
center of these 3 markers.
Z-distance evaluation is based on
estimation from the square-shape of markers, indeed large error. Any
model projected are not stable enough for an Health application !
Discussion
with Kitahara sensei. Several solutions :
Filtering
video capture, implement a low-pass filter. (-) : a delay is generated.
Adding
a second camera and filter data. (-)
: it's an expensive and not so efficient solution.
Using
a thresold to minimized the error. (+) : esay to implement.
First intention solution.
Fixing
a 3D-reference human model by stretching it to fit the markers. (+) : each marker influence only one part
of the body. The global error is minimized. (-) : difficult to
implement.
Localization of markers. Marker
position follow the OpenGL specification (i.e. matrix notation).Borrowing of a book :
"OpenGL, Programming Guide. J. Neider, T. Davis, M. Woo".
Localizing several markers
Here is a screenshot :
The camera was capturing 3 markers.
The program localize them. Positions of each markers are
processing. Relation between marker is found. A polygon is drawn
between each markers.