[ITK-users] Registration

[Dženan Zukić]

Hi Prakash,

"judge differences in 2 photographs" - without any constraints, this is a
very hard problem, and currently unsolved in science.

If you have a movement in the antenna, you can judge that by calculating a
difference to reference image. But if you sometimes have also a movement of
the camera, it becomes non-trivial to decide whether the camera has moved
somewhat or the antenna did not deploy fully.

Is this image a reference or field image? Can you attach some more so we
can see the variability among them? It is really hard to give advice when
we have to guess what kind of data you are working with.

Regards,
Dženan

On Mon, Dec 18, 2017 at 10:50 PM, Surya Prakash <ds_prakash at rediffmail.com>
wrote:

> Hello Dženan,
> Thanks for your suggestions. You asked for a sample of the data, one image
> is attached herewith. Sometimes, there may be translation / rotation
> involved in the field collected antenna image. How we can address that.
> What examples I should follow. I apologies for posting directly.
>
> Similarly I am working on another problem. How I can judge differences in
> 2 photographs acquired at different times. Translation and rotation will be
> there as these photos are acquired manually using digital cameras. I am
> looking for feature differences etc.
>
> Thanks & regards
> Prakash
>
> On Sat, 09 Dec 2017 21:09:18 +0530 Dženan Zukić wrote
> >Hi Prakash,
> what you might do is create mask of your deployed antenna and save it
> along your reference image of the antenna. Then in field, acquire a new
> image, apply the mask to both reference and field images, and then do a
> difference. Try it with several different field images with different
> lighting and states of deployment of the antenna to figure out a dependence
> of pixel-wise difference in images and whether antenna was deployed
> properly.
> Regards,Dženan
> On Sat, Dec 9, 2017 at 2:28 AM, Surya Prakash wrote:
> Hi Dženan
>
> Thank you very much for your help and guidance.
>
> The images I am working with are 2D. Camera position is fixed with respect
> to Antenna. Assuming there are no variations in foreground and background.
> Lighting / Illumination may differ from lab and field conditions.
>
> In the lab, images are acquired in still condition. Even though camera is
> fixed with respect to each other, the entire system is not stationary but
> moving.
>
> Waiting for your advise
>
>
>
> Thanks & regards
>
> Prakash
>
>
>
> On Fri, 08 Dec 2017 20:27:39 +0530 Dženan Zukić wrote
>
> >Hi Prakash,
>
> are the images you are working with 3D or 2D? If 2D, do they have variable
> lighting, variable background, variable foreground (maybe something
> obscuring the antenna)? Does the camera move relative to antenna or not?
>
> In the simplest case (camera not moving, lighting, foreground and
> background not variable) you don't need to do any registration, just
> subtract the images (reference and current) and compare difference against
> a threshold to judge whether the antenna deployed. And each added variable
> in the imaging process makes your task much harder.
>
> Is the camera just pivoting in a fixed position? You could get away with
> translation only during registration. Camera moving and pivoting, but
> pointing at the same location? You will need deformable registration,
> probably with manual initialization too. And so forth.
>
> If you want more specific advice, be more specific in your question. And
> show us some example images (reference and those being examined).
>
> And you might as well post it on discourse, more people will see it as we
> moved the discussions there.
>
> Regards,Dženan
>
> On Thu, Dec 7, 2017 at 11:01 PM, Surya Prakash wrote:
>
> Hi Dženan,
>
>
>
> Thanks for the reply. Earlier I was using ITK and after a long gap again I
> required it. Actually I am planning to use ITK for whether an antenna is
> deployed properly or not. I have antenna images (reference). I want to
> match the image of deployed antenna with the reference images to judge
> whether it is deployed properly or not. There are two things.
>
>
>
> 1. If needed I should be able to register my present image with reference
> image without using any control points (if possible)
>
>
>
> 2. Automatic change / difference between the images.
>
>
>
> As the example list is too vast can you please suggest what will be the
> right choice for me.
>
>
>
>
>
>
>
> Thanks & regards
>
>
>
> Prakash
>
>
>
>
>
>
>
>
>
>
>
> On Thu, 07 Dec 2017 20:18:17 +0530 Dženan Zukić wrote
>
>
>
> >Hi Surya,
>
>
>
> there are plenty of example for registration. You could take a look at the
> ones on the wiki.
>
>
>
> Regards,Dženan
>
>
>
> On Thu, Dec 7, 2017 at 2:09 AM, Surya Prakash wrote:
>
>
>
> Hi all
>
>
>
>
>
>
>
> I have some reference images acquired previously. Current images I want to
> register with those images. What examples of ITK I should follow
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Thanks & regards
>
>
>
>
>
>
>
> Prakash
>
>
>
>
>
>
>
> The ITK community is transitioning from this mailing list to
> discourse.itk.org. Please join us there!
>
>
>
>
>
>
>
> ________________________________
>
>
>
>
>
>
>
> Powered by www.kitware.com
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Visit other Kitware open-source projects at
>
>
>
>
>
>
>
> http://www.kitware.com/opensource/opensource.html
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Kitware offers ITK Training Courses, for more information visit:
>
>
>
>
>
>
>
> http://www.kitware.com/products/protraining.php
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Please keep messages on-topic and check the ITK FAQ at:
>
>
>
>
>
>
>
> http://www.itk.org/Wiki/ITK_FAQ
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Follow this link to subscribe/unsubscribe:
>
>
>
>
>
>
>
> http://public.kitware.com/mailman/listinfo/insight-users
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://itk.org/pipermail/insight-users/attachments/20171219/0340b493/attachment.html>