13 Apr
2013
13 Apr
'13
9:43 a.m.
On 13/04/13 00:42, Alexander Brock wrote:
- The original fitting method fits the curve to a set of points
obtained by taking 10 samples from each segment. This is a problem if the segments have very different lengths, the influence of very small parts of the original shape is way too big. I could handle this by multiplying each term by the length of the corresponding segment
Turns out this doesn't help, see attached example. I will try and get a evenly distributed set of points on the original curve.
Alexander