Greetings, I'm in the process of implementing an imaging system for my lab. The setup consists of three CCD camera's equally spaced around an excised perfused rabbit heart located in the center (i.e., the cameras are 120 degree from each other around the heart, pointing at the heart).
Since this is a stereo imaging problem, I need to come up with an appropriate calibration target (and target pattern) for the system. My initial idea is to use a hexogonal object, such that 2 planar faces would be facing each camera. However, there are a few techincal issues:
(1) In our field, researchers either use CCD cameras or photodiode arrays, with a trade-off between spatial and temporal resolution. In our case, we are using CCDs but to get the frame rates we want, the CCD will be imaging a 128 x 128 pixel area. This means that the calibration target pattern (whatever it is) has to be resolvable at a fairly low resolution.
(2) On the other hand, the CCDs will be imaging in darkened conditions, capturing florescence in the orange spectrum. I've been able to get pretty good contrast on a pattern as long as it's colored orange and black. Isolating the target from the background shouldn't be too hard either.
Our previous system consisted of one camera and two mirrors. In that case, I used a cubic shape with a grid pattern. For this one, I'm considering using a set of circular points for the pattern. Any recommendations on the pros and cons of lines vs. circles for calibration?
If my choice of a hexagonal target is a bad one, I'm open to suggestions. I should be able machine any shape I want.
Any ideas? -Mark P.S. As an aside, if anyone knows of robust MATLAB feature extraction programs (either lines or ellipses), I'd love to know.