How to remove noise from contrast

[Mcenroe_ng_teow_ee]

good morning fellow users,

RE: I need to calculate the area of the particles in my sample but is super low contrast

I have a sample of several particles that has extremely low contrast with the background. so after changing it to 8 bit and thresholding the sample. I can see the particle but when I analyze the sample it has noise from the thresholding. please guide me on how to remove the noise or alternative methods. thank you!

after threshold got a lot of small particles.
the circled dark spots is the one i want and need to calculate

need to remove the noise from the threshold

please guide!

thank you!

regards
mcenroe ng
mcenroeng@gmail.com

[Bio7]

In your case it makes sense to send the original image to the forum.
I think you already lost a great deal of information by converting it to 8-bit.

There are other strategies which might work better with the original data which has to be inspected first (hist., etc.) before thresholding.

[anon96376101]

Good day,

in any case you need some kind of background correction and I suggest to use the plugin “Polynomial Shading Corrector”.

With this plugin installed the following ImageJ-macro may get you started:

run("32-bit");
run("Polynomial Shading Corrector", "degree_x=2 degree_y=2 regularization=2");
run("Median...", "radius=5");
setAutoThreshold("Triangle");

Paste the above macro code to an empty macro window (Plugins >> New >> Macro) and run it.

You may try to lower the upper threshold a bit …

HTH

Herbie

[Mcenroe_ng_teow_ee]

@Bio7
attached is the orginal sample. once you threshold to around 220 you can start to see the dark dots. please kindly guide me along cause i am new to image J.

thanks a lot in advance.

[Bio7]

Is this really the original image? *.jpeg is always a bad idea because of the compression artifacts.

However if I split the channels (Image->Color->Split Channels), select the red channel and apply auto contrast to the image (Image->Adjust->Brightness/Contrast->Auto) I got this:

So this result is a little improvement over your conversion because we can filter the noise, e.g. with a median or gausian filter to remove the noise(Process->Filters):

You can play a little bit with different filters (and settings) to remove the noise in the image (gaussian, median) or apply other noise removal strategies to improve the output.

It will be however hard to automatically count the particles because even with the non-expert human eye for this image it is hard to identify the objects of interest.

[Mcenroe_ng_teow_ee]

will this image helps? but i only need the center white part image
@Bio7

[Bio7]

Is this a camera picture or what do we see? How was this image generated? What methods did you apply to generate the already posted different images?

This is again a *.jpeg so try to apply the methods of my previous post.

[Mcenroe_ng_teow_ee]

erm… JPG and JPEG are the same? yes, i took it from my hand-phone. I use image J or snipping tool to crop and save. i have tried all of the method but is not getting the results that i am looking for.

thanks a lot for helping.

[Mcenroe_ng_teow_ee]

Hi,

thank you for your help. hmm… i still cant get the background correction that i am looking for.

thanks a lot for your time

regards
mcenroe ng
mcenroeng@gmail.com

[anon96376101]

Here is an attempt with the built-in background correction:

run("Set Measurements...", "area redirect=None decimal=3");
run("32-bit");
run("Subtract Background...", "rolling=1000 light");
run("Median...", "radius=5");
setAutoThreshold("Triangle");
setOption("BlackBackground", false);
run("Convert to Mask");
run("Analyze Particles...", "size=150-Infinity show=Outlines display clear add");

Paste the above macro code to an empty macro window (Plugins >> New >> Macro) and run it.

With the given setting, 25 ROIs are found

with the following areas:

Taking images with a smart-phone camera for scientific purposes is an absolute no-no!
These cameras do a lot of obscure processing and the gamma is not defined. Furthermore, JPG = JPEG introduces artifacts due to lossy compression.

In short: Stay away!

Regards

Herbie