Substract diffuse reflection....

[Liisa Palomaa]

Hi..

I have just start to use ImageJ, so I have no idea if I can do what I
want. I hope someone knows if it is possible and how. I have some
images- that I want to substract the diffuse reflection from and only
have the specular reflection in the final image. The images is a
result from scanning some boards. Here is one of the images:

http://img164.imageshack.us/img164/5372/t2img1vw6.png

The "black hole" is missing data.....

The reason why I want to substract the diffuse reflection is that I
think it is easier to be able to see the fiber orientation in the
wood.

Thanks in advance

Liisa

[Till Bretschneider]

Hi Liisa,

try Process->FFT->Bandpass Filter with
Filter Large Structures down to 10
Filter Small Structures up to 0

This seems to work well apart from the region close to the hole...

Cheers,

Till


Liisa Palomaa wrote:

> Hi..
>
> I have just start to use ImageJ, so I have no idea if I can do what I
> want. I hope someone knows if it is possible and how. I have some
> images- that I want to substract the diffuse reflection from and only
> have the specular reflection in the final image. The images is a
> result from scanning some boards. Here is one of the images:
>
> http://img164.imageshack.us/img164/5372/t2img1vw6.png
>
> The "black hole" is missing data.....
>
> The reason why I want to substract the diffuse reflection is that I
> think it is easier to be able to see the fiber orientation in the
> wood.
>
> Thanks in advance
>
> Liisa
>
>

--
Dr. Till Bretschneider
AG Celldynamics (Gerisch)
Max-Planck-Institut fuer Biochemie
Am Klopferspitz 18a
D-82152 Martinsried, Germany

Tel: [phone]
Fax: [phone]
E-mail: [hidden email]
WWW: http://www.till-bretschneider.de

[Till Bretschneider]

Liisa,

option #2: you can blur your image repeatedly by using Process->Smooth
until you see no more fine structures in the background.
Then dubsract the smoothed image from the original using Process->Image
Calculator

Till

Liisa Palomaa wrote:

> Hi..
>
> I have just start to use ImageJ, so I have no idea if I can do what I
> want. I hope someone knows if it is possible and how. I have some
> images- that I want to substract the diffuse reflection from and only
> have the specular reflection in the final image. The images is a
> result from scanning some boards. Here is one of the images:
>
> http://img164.imageshack.us/img164/5372/t2img1vw6.png
>
> The "black hole" is missing data.....
>
> The reason why I want to substract the diffuse reflection is that I
> think it is easier to be able to see the fiber orientation in the
> wood.
>
> Thanks in advance
>
> Liisa
>
>

--
Dr. Till Bretschneider
AG Celldynamics (Gerisch)
Max-Planck-Institut fuer Biochemie
Am Klopferspitz 18a
D-82152 Martinsried, Germany

Tel: [phone]
Fax: [phone]
E-mail: [hidden email]
WWW: http://www.till-bretschneider.de

[Dimiter Prodanov]

>Hi,

In princliple you can use the Subtract background command but you
have to investigate the best window setting.
Similar result you can gat also by subtracting sufficiently blurred image
from the source image but again you have to asjust the kernel width parameter.

best regards



_______________________________________________________________________
Dr Dimiter Prodanov, MD, Ph.D.

Neural Engineering Rehabilitation Laboratory
(Laboratoire de Génie de la Réhabilitation Neurale)
Département de Physiologie et Pharmacologie
Université catholique de Louvain
Avenue Hippocrate, 54
POBox UCL-5446 / B-1200 Bruxelles -Belgique-
Phone: [phone]
Fax: [phone]

http://www.md.ucl.ac.be/gren

[Liisa Palomaa]

Hi..

Thanks, but it didn't work. If I first use smooth and then substract
the smoothed image from the original- I end up with a black
image......

//Liisa

[Kenneth Sloan-2]

On Oct 25, 2006, at 3:24 PM, Liisa Palomaa wrote:

> Hi..
>
> Thanks, but it didn't work. If I first use smooth and then substract
> the smoothed image from the original- I end up with a black
> image......
>
> //Liisa

My version of Unsharp Masking (which is, I think, what you are trying  
to do),
computes a weighted average of the original and the smoothed image:

Result = a*Original + b*Smoothed;   /* a + b = 1.0 */

or, more explicitly:

Result = a*Original + (1.0-a)*Smoothed;

This gives you a complete family of possible results, as you vary a.
Finally, be sure to clip the Result against 0.0 and 1.0, to avoid  
solarization
effects (or...not, if you like them - but I suspect you won't like  
them).

If your implementation looks like:

Result = Original - Smoothed;

then you will, indeed, get a very dark image.  for your application, you
should start with:

Result = 2.0*Original - 1.0*Smoothed;

and then explore from there.

For extra credit only:  instead of clipping against 0.0 and 1.0, you  
might
instead think about shrinking the range of Result values around 0.5  
if the
range of Results is too big (or expanding about 0.5 if the range is  
too narrow).
But, in my experiece, clipping is adequate.

--
Kenneth Sloan                                [hidden email]
Computer and Information Sciences                    [phone]
University of Alabama at Birmingham              FAX [phone]
Birmingham, AL 35294-1170              http://www.cis.uab.edu/sloan/

[Liisa Palomaa]

Hi...

I don't know if it's works, because I didn't understand how I can do
it. Where do you do this : Result = a*Original + (1.0-a)*Smoothed; ?

The thing I want to do is to separate the diffuse and specular
reflection.... This report is a good explanation of what I want.
http://www.cvl.iis.u-tokyo.ac.jp/papers/all/651.pdf
Is your "thing" doing the same?

The problem is that I don't know if I can use it for grey-scale images.

//Liisa

[Klughammer GmbH]

Hi Liisa,

I tried the described method and it worked quite well. But you should
start with 1.2*Original.
2.0*Original was too much for me.

You can send me an image and I'll try it for you.

Anna


Liisa Palomaa schrieb:

> Hi...
>
> I don't know if it's works, because I didn't understand how I can do
> it. Where do you do this : Result = a*Original + (1.0-a)*Smoothed; ?
>
> The thing I want to do is to separate the diffuse and specular
> reflection.... This report is a good explanation of what I want.
> http://www.cvl.iis.u-tokyo.ac.jp/papers/all/651.pdf
> Is your "thing" doing the same?
>
> The problem is that I don't know if I can use it for grey-scale images.
>
> //Liisa
>

[Liisa Palomaa]

Hi!

I have an idea how I can take away the diffuse reflection. But I need
some help, please.I attached some images so it is easier to understand
what I want to do.

http://img100.imageshack.us/img100/4790/t2img20kp4.png

I want to substract "light intesity" that is the same as the line in
the image- from all the points in the image. The line doesn't show the
exakt place- but I want to substract the light intesity- that is same
as for a line from the center of the image and up to the top of the
image or/and the center of the image to the bottom af the image. If I
do a plot profile from the top of the image to the bottom ( in the
midle in x-direction) I get the following plot:

http://img86.imageshack.us/img86/5053/plotoft2img1gj9.png

I think it is a good idea to use a median filter with radius 5 on the
plot profile- to make it more smooth, before subtracting the light
intesity from the original image.

The reason why  want to do this way - is that I am interested of the
light scattering in the horizontal direction........

 I am not sure if it will work- because the plot profile should be a
Gaussian distrubuted-(the same on both sides of the midle) when I make
the selection in the midle- but it seems like the black area- is not
always in the midle for all the images. But maybe it is good enough.
It might happen that is best to try to find the position in the
x-direction where the plot profile is Gaussian distributed..... And
then make a picture of it's light intensity-0 and then substract it
from the original image.

I would really appreciate if someone can help me and tell me if this
is possible or not.
I attached some images if someone want to try.

http://img132.imageshack.us/img132/6105/t2img100jl2.png
http://img132.imageshack.us/img132/4930/t2img1cb9.png

Thanks in advance

Liisa

[Liisa Palomaa]

Hi again...

I forgot to describe that "the light intensity" I want to substract
from the original image- is a line from the top to the center of the
image. Then I want to create an image with a circle that has the same
radiator that the line in the original image- and the same light
intensity as the line..... So the circle should be dark in the center.
Is that possible?

Thanks in advance

//Liisa

[Liisa Palomaa]

Hi...

Sorry.... I was a little bit unclear before.

The line I want to substract is a line- that I select with the
"straight line selection tool"....... So it doesn't have a constant
greyscale value. The selection should be at the position x=192,y=0 and
x=192, y=192)Then I do a plot profile and use a median filter to
smooth the plot profile. After that I want to create a new image with
a radius that is the same as the line (192 pixels) and have the same
light intesity as the line. So the circle should be dark in the midle.
 The last step is just to substract the circle image from the original
image.

//Liisa

[Liisa Palomaa]

Hi again....

I will try to explain one more time- because it is hard to understand
what I want to do....

After I have made my selection (just a straight line selection in the
original image with the position x1=192, y1=0 and x2=192 and y2=192) I
just make a plot profile for the selection. Then I use a median filter
to make it more smooth. Then I choose list- for th plot profile- and
get a list with the distance and pixelvalues.

Now to my problem- I want to create a new image with a circle with
radius 192(the same as for the selection) And have the same light
intensity for the circle as for the selection.

If it is not possible to do this kind of circles- maybe it is possible
to do 192 circles instead- So first do a circle with radius 192 and
the light intensity that the point x1=192 and y1=0 has, and then do a
next circle with radius 191 and the same light intensity that point
x=192 and y=1, and so on until the center of the circle x=192 y=192-
with a radius of 1 pixel(point).

After that I just want to substract that circle from the original image.

Does this make sense? Does anyone understand what I mean?

Sorry for posting so many mails to this list. But I really need to
solve my problem- so please if you know hao to do, let me know.

Thanks
//
Liisa