Reconstructing Curves from Points and Tangents

Reconstructing a finite set of curves from an unordered set of sample points is a well studied topic. There has been less effort that considers how much better the reconstruction can be if tangential information is given as well. We show that if curves are separated from each other by a distance D, then the sampling rate need only be O(sqrt(D)) for error-free reconstruction. For the case of point data alone, O(D) sampling is required.