VIDEO REGISTRATION: KEY CHALLENGES AND THE
POTENTIAL IMPACT OF THEIR SOLUTION TO THE FIELD OF
COMPUTER VISION
 
 
1. What have been accomplished in video frame-to-frame registration in the context of mosaics, 
 panorama, etc, and what are the new challenges?
 
2. What is the role of video registration in object-based segmentation of video captured by the 
 moving camera, and how far are we from automatic segmentation of video objects for an arbitrary
 scene in the context of MPEG-4?
 
3. Does Engineering Support Data (Telemetry) like DEM, camera orientation and location really help
 in registration of video frame with the reference image? What are the hard problems in this area?
 
4. Is registration of 3-D data with 3-D data (CT with MR) easier than registration of video (2-D)
 data with 3-D data (overly of video on CT)? Or vice versa?
 
5. Is wide base-line stereo problem solved now?
 
6. Does correlation still play an important role in registration? Has any thing new happened in
 correlation during the last 50 years?
 
7. Is video registration harder than visual tracking? Why or why not?
 
8. What is the role of image features in video registration?
 
9. What are the next most important problems which need to be solved in video registration.
 
10. What have been the most successful solutions so far and the most successful approaches?