What's new

Distinguish dark brown from black/blue in grayscale mode


annetalsma

New Member
Messages
4
Likes
2
First timer here, help would be much appreciated!
For my research I need to distinguish blue/black dye from dark brown (soil in this case). I use Photoshop CS5
This is an example of a photo after geometric distortion and the use of lasso tool:
vrwt3r.jpg
Then I have adjusted contrast/brightness/hue/saturation with sometimes black/white option etc before using grayscale and I get this:
neci8m.jpg
If you compare the second picture with the first you see that there are some dark brown parts also black/gray while I want them white or at least a totally different pixel number than blue and black. What can I change in my steps to make sure this doesn't happen, without changing the picture too much.
 

Tom Mann

Guru
Messages
7,223
Likes
4,342
Why don't you try a hue/saturation adjustment layer with settings something similar to this:

settings-hue_sat-adjustment_layer.jpg

This will produce an RGB image such as seen in the attached image. From there, you can covert it to grayscale, threshold it, or whatever else you want to do with it.

Tom M
 

Attachments

Tom Mann

Guru
Messages
7,223
Likes
4,342
Should you want to further exaggerate the differences between the brown of the soil and everything else, just crank up the remaining colors in the hue/sat adjustment layer I discussed in the previous post. This, as well as lots of other possibilities are almost instantly available using the hue/sat adjustment layer.

Tom M
 

Attachments

annetalsma

New Member
Messages
4
Likes
2
Should you want to further exaggerate the differences between the brown of the soil and everything else, just crank up the remaining colors in the hue/sat adjustment layer I discussed in the previous post. This, as well as lots of other possibilities are almost instantly available using the hue/sat adjustment layer.Tom M
Wow Tom, thanks a lot. It really works well. I think I underestimated the power of the hue/saturation tool a bit (only used the Master and then went on to 'selective color' tool). With your second photo I can get it perfectly into grayscale with the blue as total black pixels and the brown as gray pixels which I can work perfectly with. Really helpful as I need to do this for 60 pictures! Much appreciated.
 

Tom Mann

Guru
Messages
7,223
Likes
4,342
You are quite welcome. I'm glad we could help. Don't hesitate if you have any other questions.

BTW, just a general comment for future problems you might run into: Problems like this one fall into the category of "quantitative image processing". You basically got lucky that there happened to be a PS solution to this particular problem, but most of the time, Photoshop is a very, very poor choice for quantitative work. It's basically for artists and lovers, not scientists, LOL.

For quantitative image processing problems of any complexity, the go-to software for people in the image processing field is the Image Processing Toolbox of Matlab (by MathWorks).

Another choice is ImageJ by Wayne Rasband at NIH. This is much less powerful than Matlab, but a lot of biologists have developed free add-on's for ImageJ that will solve many common problems with either no, or hardly any programming required. If you think you'll run into such problems in the future, I strongly recommend you take a look at either of these two programs.

Best regards,

Tom M
 

annetalsma

New Member
Messages
4
Likes
2
Thanks for the tips. I'm not known with a lot of image processors and there is not a lot of scientists who have done my kind of research before. Basically chose Photoshop because I figured it would be user friendly and a method to replicate easily. I just checked out ImageJ and it looks pretty good, the grayscale option seems to be a bit more accurate than the one of PS somehow. However, I didn't find an add on which would solve my problem quickly (at this moment). Matlab I'm using further down in my analysis but not for pictures. Good to know there are some other options which I may use in future endeavors;)
 
Last edited:

Top