Le 25/02/2014 09:19, Agustin Lobo a écrit :
Thanks for the fast answer.
So I understand that:
1. Monteverdi2 Segmentation/Segmentation option Mean Shift uses all bands
Yes.
2. Monteverdi2 Segmentation/Segmentation option Edison Mean Shift uses
the 1st band only
Not exactly. From what I understand, it uses all bands but performs some
approximation in pixel look-up, where it only uses the first band, to
gain spped.
3. Monteverdi2 Segmentation/Segmentation option Mean Shift is the same than
Monteverdi2 Image Filtering/Mean Shift Filtering + the 3 additional
steps in Monteverdi2 Segmentation
No. Filtering part of the MeanShift algorithm is the same in both cases,
but the additional steps slightly modify the segmentation part of the
MeanShift algorithm to guarantee exact result when doing tile processing
(which the MeanShift segmentation filter can not guarantee). We have a
paper under review to explain this, I will let you know when it is ready.
Regards,
Julien
Correct?
Agus
On Tue, Feb 25, 2014 at 9:01 AM, Julien Michel <[email protected]> wrote:
Hi,
Jonathan was only talking about an optimization for speed-up made in Edison
library. The plain MeanShift implementation takes into account all bands and
does not depend on the band order.
So if you use -filter meanshift as argument to the Segmentation application,
you will get this behaviour.
Regards,
Julien
Le 25/02/2014 08:57, Agustin Lobo a écrit :
Hi!
How many bands are actualy being considered in the Mean Shift segmentation
process? According to the message by Jonathan Guinet I reproduce, only
1 band is used, but
I'm not sure if this is the case for all the process or just for part of
it.
Thanks
Agus
""""""""
Jonathan Guinet [email protected] Unsubscribe
10/21/13
to otb-users
Hy Eleni
you underline an interesting point in a previous message :
"
I notice that if I change the sequence of the bands ( ie from
RGBLIDAR to LIDARRGB ) the output clustered image is different
Is this reasonable?
"
this behavior is due to a processing optimization used by default when
wrraping EDISON (the MED_SPEEDUP algorithm) :
The bucket mode.explained in following reference :
Mean Shift Segmentation Evaluation of Optimization
Techniques Jens N. Kaftan and Andr`e A. Bell and Til Aach
To reduce exploration space a 3D hypercube is used the bucket. this
bucket is composed of spatial ranges : X,Y (the two first
dimensions) and dynamic range (the third dimension) i.e. the first
band of data. thus Only the first band of data is taken into
consideration to reduce exploration
Thus the sequence of band gives differents results.
"""""
--
Julien MICHEL
CNES - DCT/SI/AP - BPI 1219
18, avenue Edouard Belin
31401 Toulouse Cedex 09 - France
Tel: +33 561 282 894 - Fax: +33 561 283 109
--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html
You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
--- You received this message because you are subscribed to the Google
Groups "otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
--
Julien MICHEL
CNES - DCT/SI/AP - BPI 1219
18, avenue Edouard Belin
31401 Toulouse Cedex 09 - France
Tel: +33 561 282 894 - Fax: +33 561 283 109
--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html
You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
You received this message because you are subscribed to the Google Groups "otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
