Counting Barnacles on a panel

[Emiliano Pinori]

Hi,
I'm working on antifouling marine paints and had lots of panels to analyze. I found ImageJ a very helpfull tool. Anyway when it cames to the more adult and close togheter barnacles individuals on my panels i had experienced some difficoults in using my very simple macro to count for Barnacle at the ROI area.
I use a subtcract background, followed by set scale, rectangular selection find maxima within tollerance, and finally analyze Particles.

Now the problem is that I will lose a lot of adult barnacles in my results becouse they will never become not even 0.5 circular after find maxima due to the close presence of ather smaller barnacle. Or becouse the edge do not are at the same brightness all over the preimeter. How can i improve the accurancy without having too much false positives?

I don´think I asked my question in a understandable way, so i put a photo of wath i have to analyze and hope someone understand me anyway.

I need counts and area to have a % difference between Prov and Control side (just a 9 square cm ROI of each side)

http://s943.photobucket.com/albums/ad273/zapatainsvezia/Barnacles%20Exemp/

thank you Emiliano Pinori
Ph.D. student
Gothemburg University (Sweden)

[Raghavan, Srinivasan (GE, Research)]

Hi All,

Is it possible to convert a 32 bit image to a 8 bit image using ImageJ.

Thanks
Srini

[Gabriel Landini]

On Friday 30 October 2009  09:34:04 Raghavan, Srinivasan (GE, Research) wrote:
> Is it possible to convert a 32 bit image to a 8 bit image using ImageJ.

Is it possible for you to read the documentation?

G

[Joachim Wesner]

Hi,

?????? This is confusing ??????

The size in bits for a pixel is not what one would call "bit rate" !!?

Whatt you are describing is very easily done with "Image/Type/.." which can
also be called from a macro (and then you might need to reave your image),
but I wonder if your are actually meaning something totally different
because the former should be known even to absolute ImageJ beginners....
;-(


Joachim




                                                                           
             "Raghavan,                                                    
             Srinivasan (GE,                                              
             Research)"                                                 An
             <srinivasan.ragha          [hidden email]                
             [hidden email]>                                             Kopie
             Gesendet von:                                                
             ImageJ Interest                                         Thema
             Group                      Convert the bit rate              
             <[hidden email].                                            
             GOV>                                                          
                                                                           
                                                                           
             30.10.2009 08:54                                              
                                                                           
                                                                           
              Bitte antworten                                              
                    an                                                    
              ImageJ Interest                                              
                   Group                                                  
             <[hidden email].                                            
                   GOV>                                                    
                                                                           
                                                                           




Hi All,

Is it possible to convert a 32 bit image to a 8 bit image using ImageJ.

Thanks
Srini



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________

[Michael Schmid]

Hi Emiliano,

your image looks pretty difficult to process (one might even call it
hopeless). If you want to have any chance, make sure that there are no
overexposed features (in your test images, some of the brighter ones are
overexposed)

Some random thoughts:

- First of all, better illumination, avoiding reflections from the
background. (The number one rule in image processing!)

- Removing 'salt' noise (reflections from small water drops?) by
Process>Noise>Remove outliers (radius=1 seems to be a good compromise that
does not eliminate your small barnacles)

- Playing with Fourier custom filter to get rid of the vertical lines (the
filter should be very smooth, use Gaussian Blur on the filter).

- Convert to 32 bit and subtract an image created by "edge detection" from
the original.

- Do all the adult ones have a dark spot in the center? Then you could use
"Find Maxma" with bright background to search for these minima. If you
select "Maxima within tolerance" for the output, you could use Analyze
particles to get rid of unwanted ones (size, circularity); also you can
combine it with the thresholded original to get only minima above a
certain brightness.

Hope this helps a bit,

Michael
_______________________________________________________________________

On Thu, October 29, 2009 15:24, Emiliano Pinori wrote:

> Hi,
> I'm working on antifouling marine paints and had lots of panels to
> analyze.
> I found ImageJ a very helpfull tool. Anyway when it cames to the more
> adult
> and close togheter barnacles individuals on my panels i had experienced
> some
> difficoults in using my very simple macro to count for Barnacle at the ROI
> area.
> I use a subtcract background, followed by set scale, rectangular selection
> find maxima within tollerance, and finally analyze Particles.
>
> Now the problem is that I will lose a lot of adult barnacles in my results
> becouse they will never become not even 0.5 circular after find maxima due
> to the close presence of ather smaller barnacle. Or becouse the edge do
> not
> are at the same brightness all over the preimeter. How can i improve the
> accurancy without having too much false positives?
>
> I don´think I asked my question in a understandable way, so i put a photo
> of
> wath i have to analyze and hope someone understand me anyway.
>
> I need counts and area to have a % difference between Prov and Control
> side
> (just a 9 square cm ROI of each side)
>
> http://s943.photobucket.com/albums/ad273/zapatainsvezia/Barnacles%20Exemp/
>
> thank you Emiliano Pinori
> Ph.D. student
> Gothemburg University (Sweden)
>

[Emiliano Pinori]

Tank you very much Michael,
I'm testing youre ideas, and it looks much better, the resoults are more easily validable and comparable to those obtained by manual counting. This will be inserted in my next pubblication, and many people will be happy to have some help by imageJ analisys on theirs panels. I would like to cite you in my paper, if you are agree e-mail me your name and position.

PS: Of course when i shooted those photos i was'n thinking ImageJ but manual analysis. I was on the swedish coast with no much experties in photo shooting, or knwoledge of the facilities of the marine station that was guesting me. Next year I will shoot everything in a better illumination, to get less or better none overexposed areas.

Thank again for the quick answer / Emiliano Pinori

Michael Schmid-3 wrote

Hi Emiliano,

your image looks pretty difficult to process (one might even call it
hopeless). If you want to have any chance, make sure that there are no
overexposed features (in your test images, some of the brighter ones are
overexposed)

Some random thoughts:

- First of all, better illumination, avoiding reflections from the
background. (The number one rule in image processing!)

- Removing 'salt' noise (reflections from small water drops?) by
Process>Noise>Remove outliers (radius=1 seems to be a good compromise that
does not eliminate your small barnacles)

- Playing with Fourier custom filter to get rid of the vertical lines (the
filter should be very smooth, use Gaussian Blur on the filter).

- Convert to 32 bit and subtract an image created by "edge detection" from
the original.

- Do all the adult ones have a dark spot in the center? Then you could use
"Find Maxma" with bright background to search for these minima. If you
select "Maxima within tolerance" for the output, you could use Analyze
particles to get rid of unwanted ones (size, circularity); also you can
combine it with the thresholded original to get only minima above a
certain brightness.

Hope this helps a bit,

Michael
_______________________________________________________________________

On Thu, October 29, 2009 15:24, Emiliano Pinori wrote:
> Hi,
> I'm working on antifouling marine paints and had lots of panels to
> analyze.
> I found ImageJ a very helpfull tool. Anyway when it cames to the more
> adult
> and close togheter barnacles individuals on my panels i had experienced
> some
> difficoults in using my very simple macro to count for Barnacle at the ROI
> area.
> I use a subtcract background, followed by set scale, rectangular selection
> find maxima within tollerance, and finally analyze Particles.
>
> Now the problem is that I will lose a lot of adult barnacles in my results
> becouse they will never become not even 0.5 circular after find maxima due
> to the close presence of ather smaller barnacle. Or becouse the edge do
> not
> are at the same brightness all over the preimeter. How can i improve the
> accurancy without having too much false positives?
>
> I don´think I asked my question in a understandable way, so i put a photo
> of
> wath i have to analyze and hope someone understand me anyway.
>
> I need counts and area to have a % difference between Prov and Control
> side
> (just a 9 square cm ROI of each side)
>
> http://s943.photobucket.com/albums/ad273/zapatainsvezia/Barnacles%20Exemp/
>
> thank you Emiliano Pinori
> Ph.D. student
> Gothemburg University (Sweden)
>

[Robert Dougherty]

Emiliano,

I think this is a job for the Hough transform.  A way to start is to  
subtract the background, convert to 32-bit, find edges, and do an  
"adjust threshold" with limits of something like 175-743.  Now the  
little varmints are represented by reasonably well defined circles and  
arcs.  This should be perfect for the Hough transform.  The only Hough  
transform plugin that I've found so far is Hough Circles, and I can't  
seem to get it to work well in this case.  Also, the Hough Circles  
link is missing from the Plugins page. Perhaps someone else on the  
list knows how to drive it or knows of a more suitable plugin.  This  
is a rather basic capability that ImageJ should have if it does not.

Bob

On Oct 29, 2009, at 7:24 AM, Emiliano Pinori wrote:

> Hi,
> I'm working on antifouling marine paints and had lots of panels to  
> analyze.
> I found ImageJ a very helpfull tool. Anyway when it cames to the  
> more adult
> and close togheter barnacles individuals on my panels i had  
> experienced some
> difficoults in using my very simple macro to count for Barnacle at  
> the ROI
> area.
> I use a subtcract background, followed by set scale, rectangular  
> selection
> find maxima within tollerance, and finally analyze Particles.
>
> Now the problem is that I will lose a lot of adult barnacles in my  
> results
> becouse they will never become not even 0.5 circular after find  
> maxima due
> to the close presence of ather smaller barnacle. Or becouse the edge  
> do not
> are at the same brightness all over the preimeter. How can i improve  
> the
> accurancy without having too much false positives?
>
> I don´think I asked my question in a understandable way, so i put a  
> photo of
> wath i have to analyze and hope someone understand me anyway.
>
> I need counts and area to have a % difference between Prov and  
> Control side
> (just a 9 square cm ROI of each side)
>
> http://s943.photobucket.com/albums/ad273/zapatainsvezia/Barnacles%20Exemp/
>
> thank you Emiliano Pinori
> Ph.D. student
> Gothemburg University (Sweden)
> --
> View this message in context: http://n2.nabble.com/Counting-Barnacles-on-a-panel-tp3912808p3912808.html
> Sent from the ImageJ mailing list archive at Nabble.com.

Robert Dougherty, Ph.D.
President, OptiNav, Inc.
4176 148th Ave. NE
Redmond, WA 98052
[phone]
FAX [phone]
www.optinav.com
[hidden email]

[David Webster]

You probably need to do something like Process/Binary/Outline to get just
the boundary pixels before the HT. Another possibility is to do a Watershed
segmentation if the "circles" are touching.

David Webster

On Mon, Nov 2, 2009 at 5:07 PM, Robert Dougherty <[hidden email]> wrote:

> Emiliano,
>
> I think this is a job for the Hough transform.  A way to start is to
> subtract the background, convert to 32-bit, find edges, and do an "adjust
> threshold" with limits of something like 175-743.  Now the little varmints
> are represented by reasonably well defined circles and arcs.  This should be
> perfect for the Hough transform.  The only Hough transform plugin that I've
> found so far is Hough Circles, and I can't seem to get it to work well in
> this case.  Also, the Hough Circles link is missing from the Plugins page.
> Perhaps someone else on the list knows how to drive it or knows of a more
> suitable plugin.  This is a rather basic capability that ImageJ should have
> if it does not.
>
> Bob
>
>
> On Oct 29, 2009, at 7:24 AM, Emiliano Pinori wrote:
>
> Hi,
>> I'm working on antifouling marine paints and had lots of panels to
>> analyze.
>> I found ImageJ a very helpfull tool. Anyway when it cames to the more
>> adult
>> and close togheter barnacles individuals on my panels i had experienced
>> some
>> difficoults in using my very simple macro to count for Barnacle at the ROI
>> area.
>> I use a subtcract background, followed by set scale, rectangular selection
>> find maxima within tollerance, and finally analyze Particles.
>>
>> Now the problem is that I will lose a lot of adult barnacles in my results
>> becouse they will never become not even 0.5 circular after find maxima due
>> to the close presence of ather smaller barnacle. Or becouse the edge do
>> not
>> are at the same brightness all over the preimeter. How can i improve the
>> accurancy without having too much false positives?
>>
>> I don´think I asked my question in a understandable way, so i put a photo
>> of
>> wath i have to analyze and hope someone understand me anyway.
>>
>> I need counts and area to have a % difference between Prov and Control
>> side
>> (just a 9 square cm ROI of each side)
>>
>> http://s943.photobucket.com/albums/ad273/zapatainsvezia/Barnacles%20Exemp/
>>
>> thank you Emiliano Pinori
>> Ph.D. student
>> Gothemburg University (Sweden)
>> --
>> View this message in context:
>> http://n2.nabble.com/Counting-Barnacles-on-a-panel-tp3912808p3912808.html
>> Sent from the ImageJ mailing list archive at Nabble.com.
>>
>
> Robert Dougherty, Ph.D.
> President, OptiNav, Inc.
> 4176 148th Ave. NE
> Redmond, WA 98052
> [phone]
> FAX [phone]
> www.optinav.com
> [hidden email]
>