Re: Implementing SIFT algorithm

[Søren Hauberg]

man, 15 08 2011 kl. 17:32 +0530, skrev pavan padmashali:
> I'm trying to implement David Lowe's Scale Invariant Feature Transform
> code ( matlab files : siftDemoV4) on Octave. Has anyone implemented it
> on Octave yet?

What do you mean? Do you want to run David Lowe's code in Octave or do
you want to implement SIFT from scratch? From what I recall, David
Lowe's code is just a simple wrapper around a binary program so that
should easily run in Octave. If you want to re-implement SIFT in Octave,
I recommend against it: the algorithm is ridiculously hard to implement
if you want it to behave like David Lowe's implementation (you need to
get a bunch of tweaks just right or nothing works).

Søren