Skip navigation
Volorina
Currently Being Moderated

Color Select via Wand Select // Crop Via Histogram

Apr 27, 2012 1:30 PM

Tags: #cs4 #crop #histogram #select_color #magic_wand

I have two questions for CS4, which may not be possible, and I didn't find in the feature forums. 

 

1) Is there a way to do a color range selection created from all the colors that were selected from the magic wand tool.  See attached image one.

 

2) Is there a way to crop an image based on histogram?  For example if I wanted to remove all px that were above level 200 for all colors (or a specific color).  Can effect the image at all from the histogram (without an side step to something like select color range)?

 

I'm working with very large images on multiple systems (both 32 and 64 bit, all CS4).  Normally in the 50,000x30,000 -- 50,000x70,000 px range.  For the sake of argument and ease of use I could scale down but eventually need to be operating in this range.  The images are created by stitching a series of images taken under high magnification.  I would like to find ways to quantify certain regions of interested based on color, but that do not have great color separation (see image 1).  The hope would be take this information from several images and export the histogram information.  As far as cropping the histogram is concerned: the scanning process used to create the stitched images produces a lot of near white color variation that so far I've been unable to select entirely (via color range select using clusters).  I would say my best effort is only 85% of the "white" (see image 2)  I want to drop out the white px for size reduction and noise reduction when it comes to histogram statistics.

 

Also I know this is pushing the limits of intention for PS, but we tried this on three other software packages-seemingly designed for these purposes--and failed.

 

 

Pax,

Volorina

 

1:

Image1.jpg

 

2:

Image2.jpg

 
Replies
  • Currently Being Moderated
    Apr 28, 2012 9:34 AM   in reply to Volorina

    1) Is there a way to do a color range selection created from all the colors that were selected from the magic wand tool.  See attached image one.

    I know no such method, but if you start selecting with the Magic Wand Tool with »Contiguous« unchecked in the Options Bar aren’t you there already?

     

    2) Is there a way to crop an image based on histogram?  For example if I wanted to remove all px that were above level 200 for all colors (or a specific color).  Can effect the image at all from the histogram (without an side step to something like select color range)?

    I may be misunderstanding you, but could you be using »crop« but meaning »clip«?

    Could the Blend If-settings be employed for the task you have in mind?

    see screenshot (it is set to 128, though)

    clipAbove200Scr.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 29, 2012 9:18 AM   in reply to Volorina

    Let me reply first to your second question.

    "2) Is there a way to crop an image based on histogram?"

     

    As already mentioned by c.p, it appears you are using the term crop differently from conventional useage - you should look up its definition. That issue aside, the general answer is no. In Photoshop the data in a histogram are computed on the current state of the image as a result . Histogram data are not used as input to any opperation.

     

    "1) Is there a way to do a color range selection created from all the colors that were selected from the magic wand tool."

    If you meant can I use the selection (via the magic wand) as input to the color range command then the answer is no. If you meant, however, can I use that selection to find other areas with similar characteristics, then you can try clicking on Select > Similar. Starting from the selection you provided you can get the following result.

    test.jpg

     

    There are other ways to achieve similar results.

     

    I suspect you will get more help by providing us with a description of exactly what you are trying to achieve rather than asking for help on the next step in a workflow that might be flawed from the outset. How, for example, do you want to quantify your regions of interest ?  How specifically  do you define your regions of interest ? What are the tolerances on defining the regions of interest? Is the analysis to be performed on this image or on hundreds of similar images. With more detailed input you may get an answers as to whether or not Photoshop is up to the task.

     

    Paulo

     
    |
    Mark as:
  • Currently Being Moderated
    May 1, 2012 8:28 PM   in reply to Volorina

    I have read your two postings several times now, but I still have more questions than answers. I also now have a slightly stronger feeling that you may be able to get some analysis done using Photoshop.  Part of the problem is me not understanding some of the terminology you are using.

     

    Lets start simply - with image 1 in your first posting:

    In that image I see nearly white areas, I see dark burgundy areas (one of which you selected), I see dark purple areas  and I see light purple areas of various tones. I see no transparent areas. I am guessing that this is an example of a starting image on which no modifications have been made. Now, which of these various colored areas are the regions of interest ?  Physically what do each of the various colored areas represent?  

     

    Let's assume you are interested in the dark burgundy areas (since you selected one of those).  What exactly do you want to quantify about all those areas ?  Their total area ? their luminosity ?  their total perimeter ?  something else ?   You indicated a strong interest in histogram data, but from that data what analysis on it will you do ?

     

    You state "I wanted to see if we could use PS to select ROI with a small enough background that we saw an effect "   I do not understand what "effect" you are looking for and as this seems central to your question I need a better definition to determine if Photoshop can help you.  Similarly, I need to know what you mean by " blinded manual scoring"

     

    What does "a completely positive airway" mean ?

     

    Currently, my best understanding is that you are staining tissue and that areas that take the stain (various shades of burgundy) are the areas of interest, which are determined visually and somewhat imprecisely.  The process for constructing the images introduces nearly white areas, which you want to eliminate from your analysis. Finally, and most importantly, you want some measure, as yet undefined, about the stained areas - maybe relative, maybe absolute.

     

    I suspect that this is one of those situations in which understanding your problem is the biggest  issue.  Determining if Photoshop can be used and exactly how it might be used should probably go quickly.

     

    Paulo

     
    |
    Mark as:
  • Currently Being Moderated
    May 1, 2012 11:33 PM   in reply to Volorina

    Well, I must say my grip on the issue is not a lot tighter now …

    Could you please, to make it real easy, post one sample image and all the intended resulting images/evaluations you need?

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points