Re: R: R: I: HELP

Posted by Mauro Maiorca on Oct 01, 2014; 2:31pm
URL: http://imagej.273.s1.nabble.com/HELP-tp5009737p5009850.html

Hi Stefano,

it seems to me a problem easily solvable even with a simple global
threshold. If you check the intensity profiles of the structures of
interest you might see that the sealed holes have intensity profiles
higher than the average grayscale values (see attached figure1.jpg -
A) while open holes have intensity profiles lower than the average
grayscale values (see attached Fig1.jpg - B) . If you pick a threshold
value as a function of the distance from the mean value you might
solve the problem. For example, this quick and dirty Matlab script
might prove the point without even preprocessing the image (outcome is
in the attached Fig2.jpg):

A = double(imread('pores.jpg'));
ANorm = (A-mean(A(:)))/(max(A(:))-min(A(:)));
ThresholdUp = max(ANorm(:))/5;
ThresholdDown = min(ANorm(:))/5;
figure(1), imshow(mat2gray(ANorm));
figure(2), imshow(mat2gray(ANorm>ThresholdUp));
figure(3), imshow(mat2gray(ANorm<ThresholdDown));

For the preprocessing bit I would recomment a regularization filter,
either total variation or anisotropic diffusion might do the job. I'm
not sure how this can be done in imageJ, but I assume it isn't hard.

cheers,
Mauro


On Wed, Oct 1, 2014 at 2:21 PM, SureAndClean <[hidden email]> wrote:

> Hello Giorgio,
>
> no matter, I apologize since I recognize in the beginning I've not been clear.
> Coming to your question: yes, black arrows points sealed holes. The red arrows are metal oxides (iron in this case ) inclusions into the glassy layer of ceramic. These dots do not affect our investigation.
>
> This image captures about 1 mm2 of the surface whose total area is 0.4 m2 and the magnification is, for these 2 pictures 145 X.
>
> On the picture 'before' the open holes are marked with black ink that is irreversibly absorbed into them, ethanol only removes the ink not absorbed.
> The holes with circles around are partly filled.
>
> I attach an image before the marking with in kin order to let you see how a sealed pore looks like.
>
> Thanks,
>
> Best regards
>
> Stefano
>
> -----Messaggio originale-----
> Da: ImageJ Interest Group [mailto:[hidden email]] Per conto di Giorgio De Nunzio
> Inviato: mercoledì 1 ottobre 2014 12:30
> A: [hidden email]
> Oggetto: Re: R: R: I: HELP
>
> Hi Stefano,
> sorry for asking questions again, without giving any real hint, but I need to understand further.
> I think the "after" image is clear to me: I attached it again to this message, but I marked with black arrows what I think are (some of) the sealed holes: am I right?
> What are the black-ish spots I marked with red arrows? open holes?
> inhomogeneities we are not interested in?
> Does the image represent the whole sample at 150X (or so)? I am asking this because if you can see the whole sample, you can also easily compare the two images, before and after, with no registration problem, because you can use the sample borders in the two photos to coregister the images and (somehow, to be defined) subtract one from the other.
> Last question by now, in the "before" image did you cover and mask all the open holes with ink? There is perhaps one which only has a circle around? I have not understood what you did (cleaning with ethanol, what?). Sorry, but the "before" image is full of black spots and small dots that I am not able to interpret.  I'd prefer to see the clean holes with some arrows pointing to them without covering/masking. Or perhaps I am simply not understanding, sorry. The "before" image is quite different from the "Image349" you sent before, so I am quite confused.
> Thanks
> Best regards
> Giorgio
>
>
> Il 01/10/2014 11:39, SureAndClean ha scritto:
>> Hello Giorgio,
>>
>> thanks for your answer and suggestion. I will take the image with different light angle and sources.
>> Consider that the holes are not regular in their shape. I post here 2 images before and after the filling of a surface where the open and closed holes are more evident. Moreover, I outlined the open holes by marking the surface with a black permanent marker and cleaning it with Ethanol. Where there are open holes, it's visible a black mark. When the holes are sealed via a sol gel reaction you can perceive them as glassy circles.
>> Hope this is more clear,
>>
>> Thanks,
>>
>> Stefano
>>
>>
>>
>> -----Messaggio originale-----
>> Da: ImageJ Interest Group [mailto:[hidden email]] Per conto di
>> Giorgio De Nunzio
>> Inviato: mercoledì 1 ottobre 2014 09:46
>> A: [hidden email]
>> Oggetto: Re: R: I: HELP
>>
>> Ciao Stefano,
>> I am curious about your image. I think the most difficult thing would be to infer something about hole depth. Perhaps you need some sort of fixed light, with well-defined incidence angle on the sample, so that depth can somehow be related to grey levels.
>> Anyway, before all you must locate those poor-contrast spots, avoiding noise.
>> As you say that the purpose is comparison with a filled-hole image, I think you could post the corresponding image after chemical treatment, so that we can compare themspot by spot and get some ideas.
>> Are all the dark spots holes? Some of them are darker than others, some are more regular, some show a clearer disk with a central well-defined small dark circle. I think you might describe your image in more details.
>> Best regards
>> Giorgio
>>
>> --
>> Dr. Giorgio De Nunzio
>> Dipart. di Matematica e Fisica "Ennio De Giorgi", Univ. del Salento
>> INFN Sezione di Lecce ADAM (Advanced Data Analysis in Medicine) srl
>> Lecce (Italy) [hidden email]
>> [hidden email] tel [phone]-297051 fax +39 0832
>> 297100 mobile [phone]
>>
>>
>> Il 01/10/2014 08:39, SureAndClean ha scritto:
>>> Hello Volko,
>>>
>>> many thanks for your answer.
>>>
>>> I try to be more specific: I examine with an USB microscope (140 X) a ceramic flat surface.
>>> The surface presents open holes whose average diameter is less than 0.1 mm.
>>> My need is to be able, by elaborating the images I take, to create a sort of map where I outline the open holes and to get an idea of their depth (comparative purpose only).
>>> This since I would like to compare the surface after chemical treatments meant to seal the holes.
>>>
>>> Many thanks in advance,
>>>
>>> Greetings,
>>>
>>> Stefano
>>>
>>>
>>>
>>>
>>>
>>> -----Messaggio originale-----
>>> Da: ImageJ Interest Group [mailto:[hidden email]] Per conto di
>>> Volko Straub
>>> Inviato: mercoledì 1 ottobre 2014 08:00
>>> A: [hidden email]
>>> Oggetto: Re: I: HELP
>>>
>>> Hi Stefano,
>>>
>>> Could you be a bit more specific what you actually try to measure? The image is not very clear and I doubt that you will get any useful data from such an image, but I am not an expert on 'porosity'. Perhaps if you could specify more clearly what you actually try to measure and indicate the structures that you are interested in, you might stand a better chance to get an answer to your question that actually helps you.
>>>
>>> Cheers,
>>> Volko
>>>
>>>
>>> On 01/10/2014 06:40, SureAndClean wrote:
>>>> Hello,
>>>>
>>>>
>>>>
>>>> does anyone knows how can I use Image J to check the open porosity
>>>> of the surface in the attached picture?
>>>>
>>>> Thanks a million for your support,
>>>>
>>>>
>>>>
>>>> Stefano
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>
>>
>> -
>
>
> --
> Dr. Giorgio De Nunzio
> Dipart. di Matematica e Fisica "Ennio De Giorgi", Univ. del Salento INFN Sezione di Lecce ADAM (Advanced Data Analysis in Medicine) srl Lecce (Italy) [hidden email] [hidden email] tel [phone]-297051 fax [phone] mobile [phone]
>
>
>
> ---
> Questa e-mail è priva di virus e malware perché è attiva la protezione avast! Antivirus.
> http://www.avast.com
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
>
> ---
> Questa e-mail è priva di virus e malware perché è attiva la protezione avast! Antivirus.
> http://www.avast.com
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

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