Good idea, I'll do it and report back.
I'm now testing with interactive points, which should also be enlightening.
The most surprising is that
it is always the same band that gives poorer results across many images.
Agus

On Wed, Sep 2, 2015 at 2:51 PM, Guillaume Pasero
<[email protected]> wrote:
> Hi,
>
> Maybe outliers in your list of homogous points pairs between band 2 and
> band 4 ?
> Have you manually inspected your matching pairs of points ? Since there
> is almost no scaling and no rotation, you could just compute (dx,dy) for
> each matching pair, and plot it in order to find if there are wrong matches.
>
> Regards,
> Guillaume
>
> On 09/01/2015 02:21 PM, Agustin Lobo wrote:
>> I'm testing Rigid Transformation Resample to apply a model fitted out
>> of set of Homologous Points. (I'm also testing using a model derived
>> from interactive
>> points as well as Fine Registration, but that's another story).
>>
>> Interband misalignments are substantial as the images come from a
>> multicam aboard a drone. 5 bands (B,G,R,REdge,NIR).
>>
>> The procedure is
>> 1. otbcli_HomologousPointsExtraction between each band and band 4
>> (taken as reference band)
>> 2. Fit a model in R. Example of model:
>>         s               alfa                 dx           dy
>> 1 1.0029784 0.06989697  -5.819075 -3.871411
>> 2 1.0069938 0.02176913 -45.387618 -5.723231
>> 3 1.0029354 0.02747776  25.232517  7.283582
>> 4 0.0000000 0.00000000   0.000000  0.000000
>> 5 0.9994018 0.04154600   3.444670  2.325798
>>
>> So essentially no scaling, small rotation (~1 to 4 deg) and substantial 
>> shifts.
>>
>> 3. Apply model with otbcli_RigidTransformResample (2 times for each
>> image, one for the rotation then for the translation)
>>
>> After testing in 5 images, I get very good results for all bands
>> except for band 2, where the shift seems to be overestimated by
>> several pixels .
>>
>> Any idea on why?
>>
>> Thanks,
>>
>> Agus
>>
>
> --
> <www.c-s.fr>    *Guillaume PASERO*
> Ingénieur d'études et développement
> *Business Unit E-SPACE & Geo Information*
> <https://thor.si.c-s.fr/blogs/cs-blogs-business/>*- Département
> APPLICATIONS*
>
> *CS Systèmes d'Information*
> Parc de la Grande Plaine - 5, Rue Brindejonc des Moulinais - BP 15872
> 31506 Toulouse Cedex 05 - FRANCE
> +33 561 17 64 21 - [email protected]
>
> --
> --
> 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/d/optout.

-- 
-- 
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/d/optout.

Reply via email to