Pore analysis

[Lonnie Ulrich Andersen]

Hi everyone,

I am trying to figure out an easy way to do an analysis of a porous structure in 2D.

What I need is basically what you get from the thickness analysis, but instead of expanding as many diameters as can be fittet into the structure I only need it to fit the largest 5-10 diameters.

Is there an add on that can do that?

Furthermore some of the pores in the image are rather complex in shape such that one pore basically works as maybe 3 smaller ones connected by tunnels. If the thickness analysis is done one such a pore, all 10 of the largest diameters would probably fit into the same pore. Would it be possible to expand an circle in an area and then not include this area in the next search?

I hope this made sense to you guys, and that you have a solution :)


Best Regards


Lonnie Ulrich Andersen

Konsulent
Måling og Kvalitet
Produktion
Mobil [phone]
[hidden email]

<mailto:[hidden email]>Teknologisk Institut
Gregersensvej
2630 Taastrup
Telefon [phone]
http://www.teknologisk.dk<http://www.teknologisk.dk/>


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

[Jeremy Adler]

I guess you are using the distance transform (Process\Binary\Distance Map) this generates the distance from the nearest foreground pixel for the whole image.
Find the most intense pixel - this is the radius of the largest circle (pore) that can be fitted. You could then remove this circle, find the next circle and so on. To find circles that have major overlaps instead of removing the first circle from the distance map, just remove part of it (maybe just half the radius) this would leave the centres of overlapping circles with adjacent centres in place - though there will be limits to how close the centres can be - finding two circles with slightly offset centres would be a problem. But once you have found a centre you could calculate all the expected values in the distance mapped image from that centre i.e. if this circle was the only centre, and compare these to the actual values in the mapped image - differences would indicate a second overlapping centre and lead to its location.

Hough circles is another option.

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Lonnie Ulrich Andersen
Sent: den 9 november 2014 18:21
To: [hidden email]
Subject: Pore analysis

Hi everyone,

I am trying to figure out an easy way to do an analysis of a porous structure in 2D.

What I need is basically what you get from the thickness analysis, but instead of expanding as many diameters as can be fittet into the structure I only need it to fit the largest 5-10 diameters.

Is there an add on that can do that?

Furthermore some of the pores in the image are rather complex in shape such that one pore basically works as maybe 3 smaller ones connected by tunnels. If the thickness analysis is done one such a pore, all 10 of the largest diameters would probably fit into the same pore. Would it be possible to expand an circle in an area and then not include this area in the next search?

I hope this made sense to you guys, and that you have a solution :)


Best Regards


Lonnie Ulrich Andersen

Konsulent
Måling og Kvalitet
Produktion
Mobil [phone]
[hidden email]

<mailto:[hidden email]>Teknologisk Institut Gregersensvej
2630 Taastrup
Telefon [phone]
http://www.teknologisk.dk<http://www.teknologisk.dk/>


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

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

[Lonnie Ulrich Andersen]

Hi Jeremy,

Thank you for responding so quickly. I havnt really worked that much with imageJ so I probably need more help with that.

The suggestion you have would work on an image where the pores are black and object white - I have the opposite - just white blobs on the image - since I have already done a 3d analysis on the pores as if they were objects. Would that make it easier?

I could get the other image type - but then I am stuck with how to proceed after the distance map.

Thank you for trying to help 😊

Best regards

Lonnie

Sendt fra min iPhone

> Den 10/11/2014 kl. 12.04 skrev Jeremy Adler <[hidden email]>:
>
> I guess you are using the distance transform (Process\Binary\Distance Map) this generates the distance from the nearest foreground pixel for the whole image.
> Find the most intense pixel - this is the radius of the largest circle (pore) that can be fitted. You could then remove this circle, find the next circle and so on. To find circles that have major overlaps instead of removing the first circle from the distance map, just remove part of it (maybe just half the radius) this would leave the centres of overlapping circles with adjacent centres in place - though there will be limits to how close the centres can be - finding two circles with slightly offset centres would be a problem. But once you have found a centre you could calculate all the expected values in the distance mapped image from that centre i.e. if this circle was the only centre, and compare these to the actual values in the mapped image - differences would indicate a second overlapping centre and lead to its location.
>
> Hough circles is another option.
>
> -----Original Message-----
> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Lonnie Ulrich Andersen
> Sent: den 9 november 2014 18:21
> To: [hidden email]
> Subject: Pore analysis
>
> Hi everyone,
>
> I am trying to figure out an easy way to do an analysis of a porous structure in 2D.
>
> What I need is basically what you get from the thickness analysis, but instead of expanding as many diameters as can be fittet into the structure I only need it to fit the largest 5-10 diameters.
>
> Is there an add on that can do that?
>
> Furthermore some of the pores in the image are rather complex in shape such that one pore basically works as maybe 3 smaller ones connected by tunnels. If the thickness analysis is done one such a pore, all 10 of the largest diameters would probably fit into the same pore. Would it be possible to expand an circle in an area and then not include this area in the next search?
>
> I hope this made sense to you guys, and that you have a solution :)
>
>
> Best Regards
>
>
> Lonnie Ulrich Andersen
>
> Konsulent
> Måling og Kvalitet
> Produktion
> Mobil [phone]
> [hidden email]
>
> <mailto:[hidden email]>Teknologisk Institut Gregersensvej
> 2630 Taastrup
> Telefon [phone]
> http://www.teknologisk.dk<http://www.teknologisk.dk/>
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

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