-Monster Trucks Get Smart-
We're interested in developing a set of cheap, reliable, vision-based vehicles. The general idea would be to find an off-the-shelf, reasonably rugged radio-operated vehicle, to strap a small notebook to its back, and to equip it with a PCMCIA framegrabber (or to use the onboard vision available in some notebooks). The idea is then to use XVision and the FROB/FVision development environment to write interesting systems for vision-based interaction and navigation.
We are interested in updating the design and capabilities of XVision, our visual tracking system. Some initial work in this direction was done by Christopher Cantor, but much more remains to be done. Ideas for the redesign appear in a paper and a talk.
Using video of images taken of the retinal disk, develop tracking methods for computing the motion of the disk relative to the microscope, and combine this with an existing visualization system to create synchronized motion of a simulated eye.
Given images from a high-resolution endoscope, create a "mosaic" of those images; a follow-on is to register those with lower resolution images taken from an inter-ocular microscope.
Given CAT scans in the area of the knee, try extending techniques developed for 2-D affine registration under variable illumination to 3-D data. The result would be a novel method of registering "live" data to an atlas, or to a canonical bone image.
Comments to email@example.com.
Back to Top