Shadow problem

[Kashif Zeeshan]

Hi,

I have the problem of shadows in the image. When I try to count the
particles, it considers either the shadows of the particles as their part or
another particle and thus falsifies the results. Can you kindly help me in
this regard? How can I overcome this problem? Is there any plugin or easy
method to remove the shadows from the image?

Thanks a lot.

Cordially.

Kashif ZEESHAN
(00 33 6 74 90 18 10)
Doctorant, Biopesticide Group,
Laboratoire Universitaire de
Biodiversité et d'Ecologie
Microbienne (LUBEM),
6 Rue de l'Université,
29334, Quimper Cedex,
France

[David Randell]

On Thursday 12 June 2008 09:15:32 Kashif Zeeshan wrote:

> I have the problem of shadows in the image. When I try to count the
> particles, it considers either the shadows of the particles as their part
> or another particle and thus falsifies the results. Can you kindly help me
> in this regard? How can I overcome this problem? Is there any plugin or
> easy method to remove the shadows from the image?

Are the images grayscale or colour? If the latter and the objects are
reasonably homogeneous in colour you might try identifying and filtering out
the shadows by converting the image into a HSV colour-space, and manipulating
the H and V channels. See this paper for an idea how to approach this
particular problem:

http://ir.lib.u-ryukyu.ac.jp/bitstream/123456789/1489/1/No66p71.pdf

Best regards,
Dave Randell

[Kashif Zeeshan]

Hi,

Thanks for your nice suggestions. Actually, I am totally new to ImageJ and
up to now it seems me very difficult to use this. Can you kindly tell me
from where to start? The images I am using are colored imaged taken through
an ordinary camera (5M pixels) in the neon light. The liquid medium
containing aggregates was being poured in a large Petri dish and then the
photos were being taken from the above. To avoid the reflection of the light
and camera, I placed the Petri dishes in a corner of a room but due to this
now I have the shadows in the images. With the help of the replies of this
group experts, I was able to do some analysis. But ImageJ count these
shadows either as a part of the aggregates or separate aggregates. I read
the paper sent in the previous email. Thanks again for this help. But where
I will find this option of HSV color-space? I am sorry that most of the
times these terms given by the experts are difficult also and I am not able
to understand all. Can you kindly help me to sort out the problem? Where I
can find this option or how can I do this please?

Cordially.

Kashif ZEESHAN
(00 33 6 74 90 18 10)
Doctorant, Biopesticide Group,
Laboratoire Universitaire de
Biodiversité et d'Ecologie
Microbienne (LUBEM),
6 Rue de l'Université,
29334, Quimper Cedex,
France
-----Message d'origine-----
De : ImageJ Interest Group [mailto:[hidden email]] De la part de David
Randell
Envoyé : jeudi 12 juin 2008 11:22
À : [hidden email]
Objet : Re: Shadow problem

On Thursday 12 June 2008 09:15:32 Kashif Zeeshan wrote:

> I have the problem of shadows in the image. When I try to count the
> particles, it considers either the shadows of the particles as their part
> or another particle and thus falsifies the results. Can you kindly help me
> in this regard? How can I overcome this problem? Is there any plugin or
> easy method to remove the shadows from the image?

Are the images grayscale or colour? If the latter and the objects are
reasonably homogeneous in colour you might try identifying and filtering out

the shadows by converting the image into a HSV colour-space, and
manipulating
the H and V channels. See this paper for an idea how to approach this
particular problem:

http://ir.lib.u-ryukyu.ac.jp/bitstream/123456789/1489/1/No66p71.pdf

Best regards,
Dave Randell

[David Randell]

On Thursday 12 June 2008 14:01:30 Kashif Zeeshan wrote:

> Hi,
>
> Thanks for your nice suggestions. Actually, I am totally new to ImageJ and
> up to now it seems me very difficult to use this. Can you kindly tell me
> from where to start? The images I am using are colored imaged taken through
> an ordinary camera (5M pixels) in the neon light. The liquid medium
> containing aggregates was being poured in a large Petri dish and then the
> photos were being taken from the above. To avoid the reflection of the
> light and camera, I placed the Petri dishes in a corner of a room but due
> to this now I have the shadows in the images. With the help of the replies
> of this group experts, I was able to do some analysis. But ImageJ count
> these shadows either as a part of the aggregates or separate aggregates. I
> read the paper sent in the previous email. Thanks again for this help. But
> where I will find this option of HSV color-space? I am sorry that most of
> the times these terms given by the experts are difficult also and I am not
> able to understand all. Can you kindly help me to sort out the problem?
> Where I can find this option or how can I do this please?

There are several colour space related plugins one can use with ImageJ. I
would suggest you look up and install the Threshold_colour plugin by Gabriel
Landini:

http://www.dentistry.bham.ac.uk/landinig/software/software.html

Load the colour image into ImageJ and select the threshold option from the
main menu bar (the icon looks like the moon at last quarter). This will open
up a window with a set of slider bars where for the HSB option (same as HSV)
you can change the Hue, Saturation and Brightness (or Hue, Saturation and
Value for HSV). This will give you some idea of what can be done by
manipulating these channels. There are other options in ImageJ where you can
convert a colour image into a HSV stack (3 images: one for each channel), and
then extract each one in turn. If the manual manipulation of the image works,
then depending on how many images you need to process, it may be time to
consider writing a macro to process these.  For this read the Help file
associated with this plugin.

Without seeing an example image and you being a bit clearer what you want to
segment out of your images, its a bit difficult to say much more than this at
this stage.

Best regards,
Dave Randell

[Kashif Zeeshan]

Hi,

Thanks a lot for your response. I am sending you the link of my 3 photos
where I want to do the image analysis.

http://s305.photobucket.com/albums/nn227/ablaze212/

In fact, I want to count the aggregates (or particles) present in the Petri
dishes. My objective is to quantify their total number, their area, diameter
and density.

I hope I was able to explain the situation.

Thanks again. Meanwhile I am trying to do as you said.

Cordially.

Kashif ZEESHAN
(00 33 6 74 90 18 10)
Doctorant, Biopesticide Group,
Laboratoire Universitaire de
Biodiversité et d'Ecologie
Microbienne (LUBEM),
6 Rue de l'Université,
29334, Quimper Cedex,
France

-----Message d'origine-----
De : ImageJ Interest Group [mailto:[hidden email]] De la part de David
Randell
Envoyé : jeudi 12 juin 2008 18:45
À : [hidden email]
Objet : Re: Shadow problem

On Thursday 12 June 2008 14:01:30 Kashif Zeeshan wrote:
> Hi,
>
> Thanks for your nice suggestions. Actually, I am totally new to ImageJ and
> up to now it seems me very difficult to use this. Can you kindly tell me
> from where to start? The images I am using are colored imaged taken
through

> an ordinary camera (5M pixels) in the neon light. The liquid medium
> containing aggregates was being poured in a large Petri dish and then the
> photos were being taken from the above. To avoid the reflection of the
> light and camera, I placed the Petri dishes in a corner of a room but due
> to this now I have the shadows in the images. With the help of the replies
> of this group experts, I was able to do some analysis. But ImageJ count
> these shadows either as a part of the aggregates or separate aggregates. I
> read the paper sent in the previous email. Thanks again for this help. But
> where I will find this option of HSV color-space? I am sorry that most of
> the times these terms given by the experts are difficult also and I am not
> able to understand all. Can you kindly help me to sort out the problem?
> Where I can find this option or how can I do this please?

There are several colour space related plugins one can use with ImageJ. I
would suggest you look up and install the Threshold_colour plugin by Gabriel

Landini:

http://www.dentistry.bham.ac.uk/landinig/software/software.html

Load the colour image into ImageJ and select the threshold option from the
main menu bar (the icon looks like the moon at last quarter). This will open

up a window with a set of slider bars where for the HSB option (same as HSV)

you can change the Hue, Saturation and Brightness (or Hue, Saturation and
Value for HSV). This will give you some idea of what can be done by
manipulating these channels. There are other options in ImageJ where you can

convert a colour image into a HSV stack (3 images: one for each channel),
and
then extract each one in turn. If the manual manipulation of the image
works,
then depending on how many images you need to process, it may be time to
consider writing a macro to process these.  For this read the Help file
associated with this plugin.

Without seeing an example image and you being a bit clearer what you want to

segment out of your images, its a bit difficult to say much more than this
at
this stage.

Best regards,
Dave Randell

[Michael Schmid]

Hi Kashif,

one suggestion: use Image>Color to split your images into RGB.
Process>Math multiply the red channel by 1.5, then subtract 60.
Process>Image Calculator, subtract the blue minus the red channel.

It will result in an image with almost no shadows, but somewhat
noisy. You may have to smooth it before thresholding.

The better solution - like everywhere in image processing:
Half of the task is good lighting, so think about the light
first!


Michael
________________________________________________________________

On 12 Jun 2008, at 19:22, Kashif Zeeshan wrote:

> Hi,
>
> Thanks a lot for your response. I am sending you the link of my 3  
> photos
> where I want to do the image analysis.
>
> http://s305.photobucket.com/albums/nn227/ablaze212/
>
> In fact, I want to count the aggregates (or particles) present in  
> the Petri
> dishes. My objective is to quantify their total number, their area,  
> diameter
> and density.
>
> I hope I was able to explain the situation.
>
> Thanks again. Meanwhile I am trying to do as you said.
>
> Cordially.
>
> Kashif ZEESHAN
> (00 33 6 74 90 18 10)
> Doctorant, Biopesticide Group,
> Laboratoire Universitaire de
> Biodiversité et d'Ecologie
> Microbienne (LUBEM),
> 6 Rue de l'Université,
> 29334, Quimper Cedex,
> France
>
> -----Message d'origine-----
> De : ImageJ Interest Group [mailto:[hidden email]] De la part  
> de David
> Randell
> Envoyé : jeudi 12 juin 2008 18:45
> À : [hidden email]
> Objet : Re: Shadow problem
>
> On Thursday 12 June 2008 14:01:30 Kashif Zeeshan wrote:
>> Hi,
>>
>> Thanks for your nice suggestions. Actually, I am totally new to  
>> ImageJ and
>> up to now it seems me very difficult to use this. Can you kindly  
>> tell me
>> from where to start? The images I am using are colored imaged taken
> through
>> an ordinary camera (5M pixels) in the neon light. The liquid medium
>> containing aggregates was being poured in a large Petri dish and  
>> then the
>> photos were being taken from the above. To avoid the reflection of  
>> the
>> light and camera, I placed the Petri dishes in a corner of a room  
>> but due
>> to this now I have the shadows in the images. With the help of the  
>> replies
>> of this group experts, I was able to do some analysis. But ImageJ  
>> count
>> these shadows either as a part of the aggregates or separate  
>> aggregates. I
>> read the paper sent in the previous email. Thanks again for this  
>> help. But
>> where I will find this option of HSV color-space? I am sorry that  
>> most of
>> the times these terms given by the experts are difficult also and  
>> I am not
>> able to understand all. Can you kindly help me to sort out the  
>> problem?
>> Where I can find this option or how can I do this please?
>
> There are several colour space related plugins one can use with  
> ImageJ. I
> would suggest you look up and install the Threshold_colour plugin  
> by Gabriel
>
> Landini:
>
> http://www.dentistry.bham.ac.uk/landinig/software/software.html
>
> Load the colour image into ImageJ and select the threshold option  
> from the
> main menu bar (the icon looks like the moon at last quarter). This  
> will open
>
> up a window with a set of slider bars where for the HSB option  
> (same as HSV)
>
> you can change the Hue, Saturation and Brightness (or Hue,  
> Saturation and
> Value for HSV). This will give you some idea of what can be done by
> manipulating these channels. There are other options in ImageJ  
> where you can
>
> convert a colour image into a HSV stack (3 images: one for each  
> channel),
> and
> then extract each one in turn. If the manual manipulation of the image
> works,
> then depending on how many images you need to process, it may be  
> time to
> consider writing a macro to process these.  For this read the Help  
> file
> associated with this plugin.
>
> Without seeing an example image and you being a bit clearer what  
> you want to
>
> segment out of your images, its a bit difficult to say much more  
> than this
> at
> this stage.
>
> Best regards,
> Dave Randell

[ali ayata]

i don't know it would be helpfull, but have suggestions about capturing. (if it would be  possible to recapture your work..)
 
using indirect sunlight during photography would be maintain homogenous illuminated image.
 
you can also use backlight illumination with a horizontally placed negatoscope (x ray film viewer). it would be prevent shading.
 
or do not place the petri on a surface, try to hold it in the air by fixing form its edge.
 
i hope it would be helpful ...
 
ali ayata
istanbul
 


--- On Thu, 6/12/08, Kashif Zeeshan <[hidden email]> wrote:

From: Kashif Zeeshan <[hidden email]>
Subject: Re: Shadow problem
To: [hidden email]
Date: Thursday, June 12, 2008, 1:22 PM

Hi,

Thanks a lot for your response. I am sending you the link of my 3 photos
where I want to do the image analysis.

http://s305.photobucket.com/albums/nn227/ablaze212/

In fact, I want to count the aggregates (or particles) present in the Petri
dishes. My objective is to quantify their total number, their area, diameter
and density.

I hope I was able to explain the situation.

Thanks again. Meanwhile I am trying to do as you said.

Cordially.

Kashif ZEESHAN
(00 33 6 74 90 18 10)
Doctorant, Biopesticide Group,
Laboratoire Universitaire de
Biodiversité et d'Ecologie
Microbienne (LUBEM),
6 Rue de l'Université,
29334, Quimper Cedex,
France

-----Message d'origine-----
De : ImageJ Interest Group [mailto:[hidden email]] De la part de David
Randell
Envoyé : jeudi 12 juin 2008 18:45
À : [hidden email]
Objet : Re: Shadow problem

On Thursday 12 June 2008 14:01:30 Kashif Zeeshan wrote:
> Hi,
>
> Thanks for your nice suggestions. Actually, I am totally new to ImageJ and
> up to now it seems me very difficult to use this. Can you kindly tell me
> from where to start? The images I am using are colored imaged taken
through
> an ordinary camera (5M pixels) in the neon light. The liquid medium
> containing aggregates was being poured in a large Petri dish and then the
> photos were being taken from the above. To avoid the reflection of the
> light and camera, I placed the Petri dishes in a corner of a room but due
> to this now I have the shadows in the images. With the help of the replies
> of this group experts, I was able to do some analysis. But ImageJ count
> these shadows either as a part of the aggregates or separate aggregates. I
> read the paper sent in the previous email. Thanks again for this help. But
> where I will find this option of HSV color-space? I am sorry that most of
> the times these terms given by the experts are difficult also and I am not
> able to understand all. Can you kindly help me to sort out the problem?
> Where I can find this option or how can I do this please?

There are several colour space related plugins one can use with ImageJ. I
would suggest you look up and install the Threshold_colour plugin by Gabriel

Landini:

http://www.dentistry.bham.ac.uk/landinig/software/software.html

Load the colour image into ImageJ and select the threshold option from the
main menu bar (the icon looks like the moon at last quarter). This will open

up a window with a set of slider bars where for the HSB option (same as HSV)

you can change the Hue, Saturation and Brightness (or Hue, Saturation and
Value for HSV). This will give you some idea of what can be done by
manipulating these channels. There are other options in ImageJ where you can

convert a colour image into a HSV stack (3 images: one for each channel),
and
then extract each one in turn. If the manual manipulation of the image
works,
then depending on how many images you need to process, it may be time to
consider writing a macro to process these.  For this read the Help file
associated with this plugin.

Without seeing an example image and you being a bit clearer what you want to

segment out of your images, its a bit difficult to say much more than this
at
this stage.

Best regards,
Dave Randell

[Kashif Zeeshan]

Hi experts,

Its really very kind that there are a lot of experts always ready to help. I
am very thankful to all of you. Your suggestions are very valuable for me
and helping a lot. Your detailed guidelines to do the things is really
marvellous. I will follow your suggestions and will contact you in case I
have some other problems. So nice of you all.

Thanks again.

Cordially.

Kashif ZEESHAN
(00 33 6 74 90 18 10)
Doctorant, Biopesticide Group,
Laboratoire Universitaire de
Biodiversité et d'Ecologie
Microbienne (LUBEM),
6 Rue de l'Université,
29334, Quimper Cedex,
France