Finding area of a triangle shape within a complex image of a single color

[Burinok]

Hello All,
   I have been searching for a way to find the area of a piece of an object in imageJ. Unfortunately, there is no contrast difference to be used. I attach an example that I drew up in powerpoint here. The goal would be to find the area of the triangle on the left. I have not come across any solutions that could do it if the object was all of a single color. Does anyone have any suggestions?
Thanks,
Tony

[nicola giacobbe]

There are many solutions: I cannot guide you to solve it completely into  
imagej as I am not proficient on it as I would like but I could give you a  
couple hints:

1) a corner detector could find all corner on the image, it is then up to  
you to find the ones that fit the triangle you are looking for
2) a sophisticated pattern matching could find it (for an example look up  
TLD on Google)
3) other solutions depend on the specific image being analyzed (i.e. if  
the strand of the triangle were always much smaller of the triangle body  
you could ERODE the whole image, or better yet, only the colour plane  
coincident with the triangle colour, to remove the strand and then use a  
connected component labeller to recover the triangle)

Sincerely yours
  nicola giacobbe

On Sat, 20 Jul 2013 21:58:31 +0200, Burinok <[hidden email]>  
wrote:

> Hello All,
>    I have been searching for a way to find the area of a piece of an  
> object
> in imageJ. Unfortunately, there is no contrast difference to be used. I
> attach an example that I drew up in powerpoint here. The goal would be to
> find the area of the triangle on the left. I have not come across any
> solutions that could do it if the object was all of a single color. Does
> anyone have any suggestions?
> Thanks,
> Tony
>
> <http://imagej.1557.x6.nabble.com/file/n5004059/testimage.png>
>
>
>
> --
> View this message in context:  
> http://imagej.1557.x6.nabble.com/Finding-area-of-a-triangle-shape-within-a-complex-image-of-a-single-color-tp5004059.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
Using Opera's revolutionary e-mail client: http://www.opera.com/m2/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[dscho]

Hi Tony,

On Sat, 20 Jul 2013, Burinok wrote:

>    I have been searching for a way to find the area of a piece of an object
> in imageJ. Unfortunately, there is no contrast difference to be used.

In your example, there is quite a clear contrast between blue (foreground0
and white (background)... what do your *real* images look like?

> I attach an example that I drew up in powerpoint here. The goal would be
> to find the area of the triangle on the left.

Note that there is also a triangle on the right, even if its upper and
lower corners do not extend beyond the middle rectangle.

> I have not come across any solutions that could do it if the object was
> all of a single color. Does anyone have any suggestions?

Traditionally, ImageJ's strength lies in bio-medical image processing
applications, and this one strikes me more like a traditional, industrial
machine vision problem.

Having said that, if a real big (but simple) hammer fits your needs, you
could implement a custom Hough transform that identifies triangles.

Ciao,
Johannes

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Olivier Burri]

Hi there,

Maybe you could use a k-means clustering algorithm to put together all objects of the same color. Then you could easily threshold and count them.

Of course, without seeing the actual image, I cannot say whether it will work.

Best,

Oli

Olivier Burri
Engineer - Image Processing
& Software Development
EPFL - SV - PTECH - PTBIOP


-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Johannes Schindelin
Sent: lundi 22 juillet 2013 17:02
To: [hidden email]
Subject: Re: Finding area of a triangle shape within a complex image of a single color

Hi Tony,

On Sat, 20 Jul 2013, Burinok wrote:

>    I have been searching for a way to find the area of a piece of an
> object in imageJ. Unfortunately, there is no contrast difference to be used.

In your example, there is quite a clear contrast between blue (foreground0 and white (background)... what do your *real* images look like?

> I attach an example that I drew up in powerpoint here. The goal would
> be to find the area of the triangle on the left.

Note that there is also a triangle on the right, even if its upper and lower corners do not extend beyond the middle rectangle.

> I have not come across any solutions that could do it if the object
> was all of a single color. Does anyone have any suggestions?

Traditionally, ImageJ's strength lies in bio-medical image processing applications, and this one strikes me more like a traditional, industrial machine vision problem.

Having said that, if a real big (but simple) hammer fits your needs, you could implement a custom Hough transform that identifies triangles.

Ciao,
Johannes

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html