[Insight-users] Re: How to identify four corners of the image?

[Luis Ibanez]

Hi Valli,


Could you please provide more details about
the type of processing you are interested on.


by "identify corners" do you mean:



1) To find the pixels in the corners of a 2D image ?

    This first one is trivial. Simply take the
    LargestPossibleRegion of the image. the GetIndex()
    will return the first corner. GetIndex()+GetSize()
    will return the oposite corner. The remaining corners
    in 2D can be find easily by permuting x,y components.


Or do you mean



2) To find objects inside the image whose contours
    have right angles (or close to right).
    Like a computer vision task for finding
    man-made objects in a scene ?


If this is the case you may want to construct a
pipeline using a set of RecursiveGaussianImageFilters
Similar to what is done in the demo-application
Curve2DExtraction in the directory

    /Insight/Applications/Curves2DExtractor


The analysis of derivatives of the image in a
parametric space will provide you with a framework
for performing pattern-recognition.


Please let us know more details about your target.



   Thanks


    Luis



===========================

cspl wrote:

> Hi All !
> 
>  
> 
> In one of our applications we need to identify the corners of an image, 
> starting from corners identify all the pixels connected to any of the 
> four corners of the image.
> 
>  
> 
> Can any body help us how to identify the corners of the image??
> 
>  
> 
> Regards,
> 
>  
> 
> Valli
>