[Insight-users] How about use itk in remote sensing field?

Luis Ibanez luis.ibanez@kitware.com
Thu, 03 Apr 2003 05:26:07 -0500


Hi Chen,

ITK was designed with Medical Images in mind,
but the algorithms included in the toolkit are
quite generic. There is no fundamental reason
why it could not be used for remote sensing.
The tasks of registration and segmentation are
similar in remote sensing. (e.g. you will find
interesting things like curve extraction, statistical
classification, multi-modality registration,
edge-preserving smoothing, multi-resolution...).

ITK supports multi-band images thanks to the
generic programming design. It also supports
streaming in order to process very large datasets,
since medical images can easily reach the GigaByte
range too.

ITK is not dependent of FLTK at all. There is no
link between ITK and FLTK except for the fact that
some of the demo applications use FLTK for GUI
support. Other examples use Qt, and of course you
could use MFC too. Recently we are embedding ITK
into other applications that have their own GUI,
for example:

  - Analyze
  - SCIRun
  - VolView

In all these cases ITK is being encapsulated in
small modules that are called from the top
application.

You may want to take a look at the code in

   InsightApplications/VolviewPlugIns

to see how ITK components can be encapsulated
in modules.

One of the first things you may want to do when
designing modules is to define a subset of
possible image types and instantiate the modules
for these types, then use switch/case statements
inside the modules for selecting at run time the
appropriate image type. In this way your drag/drop
module is type-independent from the outside and
can still manage a variety of image types.



Regards,


   Luis



-------------------
Chen Fu wrote:
> Hi!
> I take a job in a national remote sensing institute of
> China. In this institute, we receive and produce the
> images of many satellites like LandSat(US),
> SPOT(France) and so on.
> 
> I used opendx to do some processing on large image.
> But opendx is still a linux/unix software, the port to
> windows need many other stuff like cigwin or exceed(a
> simulator of X windows).
> 
> ITK is a very good package for me. it's open source
> and platform independent. and what's more it is
> written by template, i like STL and template concept,
> that make the c++ a unique language against java or
> c#.
> 
> Now my interest is using ITK in remote sensing field.
> I hope to build a Flow design interface use pipeline.
> like the VIP(IDL), the user can drop several filter
> blocks into the canvas, draw lines between ports of
> blocks , change the text box on the blocks to tune the
> parameter of filters. Then the flow can be run, and
> user can check very result after a filter. I know itk
> application use FLTK to own a GUI, but i think most
> common users don't like programming, a drag-drop
> pattern is much easier.
> 
> I am just a coder, and know little about the
> segmentation and registration algorithm. Do you think
> these algorithm used in medical image also fit the
> satellite image and airplane photo? Is there any
> parellel computing design in ITK, for the satellite is
> very large(hundreds M or several G)?
> 
> any suggestion?
> 
> 
> 
> =====
> Remote Scensing Satellite Ground Station
> Chinese Academy of Science
> 
> _________________________________________________________
> Do You Yahoo!? 
> 雅虎通网络KTV, 随时随地免费卡拉OK~~
> http://rd.yahoo.com/mail_cn/tag/?http://cn.messenger.yahoo.com//chat/index.html
> _______________________________________________
> Insight-users mailing list
> Insight-users@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-users
>