crack length determination

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

crack length determination

negupta
Hi,

Myself Neel Gupta. I am trying to use X-Ray CT scan images of a rock core to determine the length of crack generated in it after load application. To serve the purpose of crack identification in images I used Canny edge detector. It gave me good results regarding crack position in image. But still I am not able to determine the length of the crack extending in z-direction.

For analyzing crack length in concrete, BoneJ has been advised to use but it is not giving me desired results.

Please share your views to complete such analysis.
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

Jan Eglinger
Dear Neel,

there has been a discussion recently about a similar topic on the ImageJ
forum:

http://forum.imagej.net/t/crack-analysis-in-3d-ct-results-with-fiji/876?u=imagejan

http://forum.imagej.net/search?q=crack

Hope that helps,
Jan


On 28.02.2017 00:29, negupta wrote:

> Hi,
>
> Myself Neel Gupta. I am trying to use X-Ray CT scan images of a rock core to
> determine the length of crack generated in it after load application. To
> serve the purpose of crack identification in images I used Canny edge
> detector. It gave me good results regarding crack position in image. But
> still I am not able to determine the length of the crack extending in
> z-direction.
>
> For analyzing crack length in concrete, BoneJ has been advised to use but it
> is not giving me desired results.
>
> Please share your views to complete such analysis.
>
>
>
> --
> View this message in context: http://imagej.1557.x6.nabble.com/crack-length-determination-tp5018201.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

negupta
Thanks Jan for making me aware about TrainWeka Segmentation. It is working fine.

But my next problem is to join cracks from stacks so that I can determine their length along the length of stack. With my current knowledge I can just determine the length of crack in one particular slice. When I am stacking these slices together I want to calculate the length of crack along z direction present in all slices.

Neel
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

Herbie
Dear Neel,

the problem of tracing structures in z-direction exists in various
fields, e.g. the tracing of vessels and neuronal fibers in tissue. To
date no really satisfying solution for the reliable and fully automatic
tracing in z is known.

Good luck

Herbie

:::::::::::::::::::::::::::::::::::::
Am 01.03.17 um 00:54 schrieb negupta:

> Thanks Jan for making me aware about TrainWeka Segmentation. It is
> working fine.
>
> But my next problem is to join cracks from stacks so that I can
> determine their length along the length of stack. With my current
> knowledge I can just determine the length of crack in one particular
> slice. When I am stacking these slices together I want to calculate
> the length of crack along z direction present in all slices.
>
> Neel
>
>
>
> -- View this message in context:
> http://imagej.1557.x6.nabble.com/crack-length-determination-tp5018201p5018208.html
>
>
Sent from the ImageJ mailing list archive at Nabble.com.
>
> -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

Jan Eglinger
In reply to this post by negupta
Hi Neel,

I very much recommend KNIME Image Processing [1,2] for doing
segmentation and measurements in 3D. As ImageJ(2), it is based on
ImgLib2 and therefore fully supports n-dimensional image data.

Feel free to create a new topic on the ImageJ forum if you need help
setting up a KNIME workflow.

Cheers,
Jan



[1]: http://forum.imagej.net/t/knime-image-processing/91?u=imagejan
[2]: http://knime.imagej.net


On 01.03.2017 00:54, negupta wrote:

> Thanks Jan for making me aware about TrainWeka Segmentation. It is working
> fine.
>
> But my next problem is to join cracks from stacks so that I can determine
> their length along the length of stack. With my current knowledge I can just
> determine the length of crack in one particular slice. When I am stacking
> these slices together I want to calculate the length of crack along z
> direction present in all slices.
>
> Neel
>
>
>
> --
> View this message in context: http://imagej.1557.x6.nabble.com/crack-length-determination-tp5018201p5018208.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

negupta
Thanks again for the new probable solution.

I have come up with another possible solution to solve this problem using ImageJ. But I have one more issue and if I get satisfactory result from it may be I can share it for future purpose.

So the situation is like this:
In a 8-bit image, pixel with 255 value (dark) are representing cracks in concrete. Now as a crack is a linear feature extending over some length, dark pixels are forming a linear shape. Does anybody know how to trim the corner pixels in such linear arrangement so that I can get to the middle pixel? I know I can do it visually but then it would not be accurate enough.

Please ask If you don't understand the fact. I am sharing the image so that you can understand more what I am trying to say.

Neelblack lines are crack in image
Reply | Threaded
Open this post in threaded view
|

Re: crack length determination

Michael Schmid-3
Neel wrote:

 > In a 8-bit image, pixel with 255 value (dark) are representing cracks
in concrete. Now as a crack is a linear feature extending over some
length, dark pixels are forming a linear shape. Does anybody know how to
trim the corner pixels in such linear arrangement so that I can get to
the middle pixel?


Hi Neel,

you can use Process>Find Maxima with 'light background'. For contiguous
areas with a constant pixel value, it gives you the pixel of this area
(this crack, in your case) that is closest to the center of mass of all
pixels of this crack.

You can select whether you want a point selection, an image with only
the points of relevance or a list with the coordinates in the Results table.

As it seems some cracks are not contiguous in your sample image (but
they probably are in reality), you may need some preprocessing to join
the segments. The easiest would be dilation by Process>Filters>Maximum
followed by Skeletonize.

Michael
________________________________________________________________
On 01/03/2017 19:39, negupta wrote:

> Thanks again for the new probable solution.
>
> I have come up with another possible solution to solve this problem using
> ImageJ. But I have one more issue and if I get satisfactory result from it
> may be I can share it for future purpose.
>
> So the situation is like this:
> In a 8-bit image, pixel with 255 value (dark) are representing cracks in
> concrete. Now as a crack is a linear feature extending over some length,
> dark pixels are forming a linear shape. Does anybody know how to trim the
> corner pixels in such linear arrangement so that I can get to the middle
> pixel? I know I can do it visually but then it would not be accurate enough.
>
> Please ask If you don't understand the fact. I am sharing the image so that
> you can understand more what I am trying to say.
>
> Neel
> <http://imagej.1557.x6.nabble.com/file/n5018215/Classification_result_skeleton.png>
>
>
>
> --
> View this message in context: http://imagej.1557.x6.nabble.com/crack-length-determination-tp5018201p5018215.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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