Hi everybody,
I am working on a project that involves biological
cell image analysis. Our application requires a real time image
processing system that, given the microscopic images of the
experimented cell, outputs the centroid position and orientation
of the cell. This information is then used to measure DiElectroPhoresis
characteristics of the cell.
The system contains a microscope which is focused on the cells
floating in a transparent chamber. A CCD camera grabs the microscopic
image at the rate of 30 frames/second. An image processing block is
needed to get the image and calculates the centroid of the floating
cell at the same rate. The orientation of the cell is also required
to be found. The orientation of the cell is needed to measure the
rotational speed of the cell. In our application, there are many
cases in which the cell is spinning in its place and the spinning rate
is desired.
Currently, an image processing hardware made by Matrox Company, Quebec,
Canada (Imager-AT 1987) together with a software is being used.
However, the following difficulties in our system have been encountered
so far:
- Although the spherical cells are chosen for this experiments,
the cell images obtained are not always spherical. The non-
uniform density of the inner part of the cell is the main
reason for this appearance. This appearance causes faulty
detections of the centroid position. The image analysis
system has to reconstruct the real shape of the cell before
centroid detection.
- The process is not fast enough to detect quick cell movements.
- The frames with multiple cells yield erroneous centroid
position. The system cannot distinguish between the cells.
We might have to use a faster hardware.
An alternative solution to this problem can be using Analog
Position Sensor ICs. But I do not know of the ICs available or
where to start looking for these ICs.
I would appreciate if anybody can help me or give some pointers in this
regard. If you know of anybody working on a similar project please let
me know. Please send me E_mail at the following address:
najafi at enel.ucalgary.ca
Thank you in advance.
Kazem Najafi
