
	David Suter - Reearch Projects

	SITE NAVIGATION


	Urban Scanning and Modelling
	Face Recognition
	Visual Surveillance
	Human Motion Capture and Modelling
	3D Reconstruction
	Mobile Vision Guided Robot
	Historical Film Restoration
	Image Motion Estimation
	Tracking
	Medical Imaging
	Splines

David Suter - Research Project

Fast Approximation of Splines

Splines allow one to approximate the solution to many engineering and scientific problems. In this project (and related projects), we essentially did two things: investigate a new family of splines with interesting and useful properties (PDE splines), and developed techniques to calculate the spline solutuions very rapidly. The latter promises to make the computation of these splines fast enough for real world applications. The approaches we used were based upon Fast Multipole Techniques.

Applications include: robotics, surveillance, multimedia, medical imaging etc.

Applications to Medical Imaging

For example, we used these splines to recover the 3D motion of the human heard from MRI images.

Click on the images to see movie files

Calculation of Splines Using Fast Multipole Techniques

The direct method of evaluating the PDE splines will grow rapidly with the number of control points (data points) whereas the Fast Multipole Method we developed will grow more slowly - the savings in computation can be very large when dealing with a very large number of control points.