__/ [ketabat] on Friday 18 November 2005 16:37 \__
> How Can i improve active shape models? (if the training samples are not
> changed)
I suggest you look at:
R.H. Davies and C.Twining and T.F. Cootes and C.J. Taylor, "A Minimum
Description Length Approach to Statistical Shape Modelling", IEEE
Transactions on Medical Imaging Vol.21, pp.525-537, 2002
> if we use larger shape space, are we get more faithful the shape
> reconstruction?
This depends what you mean by "faithful". A larger shape space probably means
more landmark points, which by definition ought to give you better models
with smaller residual errors.
> what's happening if we increase landmark in training image? for example
> one sampling one distance in image
> regards
> ketabat
I could not understand the lattest sentence. I guess that the answer would be
that models then become better, at the expense of complexity.
Best wishes,
Roy
--
Roy S. Schestowitz | "Pentiums melt in your PC, not in your hand"
http://Schestowitz.com | SuSE Linux | PGP-Key: 0x74572E8E
5:55pm up 15 days 13:49, 4 users, load average: 0.36, 0.43, 0.34
http://iuron.com - next generation of search paradigms
|
|
