[Insight-users] Contour extraction

[Luis Ibanez]

Hi Preben,

Well, I wouldn't call it    "wrong".

I would call it :    "Incomplete".

The Canny Edge detector still takes you half-way
to where you want to go.

Now you can take the Canny Edges from the digital
image and extract paths through them that can be
represented as poly-lines.

It is not uncommon for any image analysis task to have
to connect together multiple stages of intermediate
processing. That's indeed one of the reasons why we
use a data-pipeline architecture.


For a second stage, you may have to adapt some of
the classes that manage paths. For example:

Insight/Code/Common/

itkChainCodePath.h
itkChainCodePath.txx
itkChainCodePath2D.cxx
itkChainCodePath2D.h
itkFourierSeriesPath.h
itkFourierSeriesPath.txx
itkImageAndPathToImageFilter.h
itkImageAndPathToImageFilter.txx
itkOrthogonallyCorrected2DParametricPath.cxx
itkOrthogonallyCorrected2DParametricPath.h
itkParametricPath.cxx
itkParametricPath.h
itkParametricPath.txx
itkPath.h
itkPath.txx
itkPathAndImageToPathFilter.h
itkPathAndImageToPathFilter.txx
itkPathConstIterator.h
itkPathConstIterator.txx
itkPathFunctions.h
itkPathIterator.h
itkPathIterator.txx
itkPathSource.h
itkPathSource.txx
itkPathToPathFilter.h
itkPathToPathFilter.txx
itkPolyLineParametricPath.h
itkPolyLineParametricPath.txx



       Regards,


              Luis



------------------------------------------------------------------------------
On Fri, Mar 19, 2010 at 3:43 AM, Preben <64bit at mailme.dk> wrote:
> Hi again
>
> This is wrong.
> I want an output that is vectorized (like a set of points with 1px
> separation along the contour). My output should not be a new image.
>
> Output like:
> std::vector<std::vector<Point2D> >
>
> or some other descripter of a 2D path. E.g. a spline or something like that.
>
>
> - Preben
>
>
>
>
> Den 18-03-2010 23:49, Luis Ibanez skrev:
>>
>> Hi Preben,
>>
>> Thanks for the clarification.
>>
>> It look then like you can take advantage of the
>> Canny Edge Detector, as Sergio suggested.
>>
>> You may want to look at the example in :
>>
>>        Insight/Examples/Filtering/
>>               CannyEdgeDetectionImageFilter.cxx
>>
>>
>> As usual,
>> some parameters fine-tunning will be required
>> for your specific images.
>>
>>
>>    Regards,
>>
>>
>>          Luis
>>
>>
>>
>> -------------------------------------------------------------------------------------------------
>> On Thu, Mar 18, 2010 at 3:55 PM, Preben<64bit at mailme.dk>  wrote:
>>
>>>
>>> Hi again
>>>
>>> Well, the problem is not related to organs, vessels or anything inside
>>> human beings.
>>>
>>> I'm trying to extract arbitrary shapes of "edges" in an image: contours.
>>> I don't need the contours to be closed - they can be open - no problem.
>>> The input could e.g. be a cup where I would like to extract the sharp
>>> edges on the cup and the contours formed by the difference between the
>>> cup and the background.
>>>
>>> So, in this example, the boundaries of the cup are very interesting and
>>> the internal edges/contours of the cup are of interest. I only have one
>>> picture as an input and one picture as output.
>>>
>>>
>>> - Preben
>>>
>>>
>>> Luis Ibanez wrote:
>>>
>>>>
>>>> Hi Preben,
>>>>
>>>> Please give a description of the kind of input image that you have
>>>> and the kind of segmentation that you want to perform.
>>>>
>>>> Are you trying to extract linear structures ? (tubes, blood vessels ?)
>>>>
>>>> or
>>>>
>>>> Are you trying to extract boundaries of large organs ? (livers,
>>>> kidneys)?
>>>>
>>>>
>>>>    Please be more concise,
>>>>
>>>>    Image segmentation is a very large field....
>>>>
>>>>
>>>>
>>>>        Thanks
>>>>
>>>>
>>>>              Luis
>>>>
>>>>
>>>> -----------------------------------------------------------------
>>>> On Thu, Mar 18, 2010 at 5:51 AM, Preben<64bit at mailme.dk>  wrote:
>>>>
>>>>
>>>>>
>>>>> Hi
>>>>>
>>>>> I'm wondering if it is possible to extract all the contours of the
>>>>> image
>>>>> automatically.
>>>>>
>>>>> I've read in the document Fast Marching Minimal Path Extraction in ITK
>>>>>
>>>>> (http://ezctwsgi.maxturnheim.webfactional.com/site_media_ezct/projects/matsur/jell/jell_matsur_aou_12_hea_dr_10_d.pdf),
>>>>> that it is possible to extract a path given an image, a socalled speed
>>>>> function and a start point and end point. However, I'm interested in
>>>>> all of
>>>>> the contours of the image - and the start point is not really that
>>>>> interesting.
>>>>>
>>>>> Is this possible?
>>>>>
>>>>>
>>>>> Best regards
>>>>> Preben
>>>>>
>>>>> _____________________________________
>>>>> 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.html
>>>>>
>>>>> 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://www.itk.org/mailman/listinfo/insight-users
>>>>>
>>>>>
>>>>>
>>>
>>>
>
>