Legacy ImageMagick Discussions Archive

Hi,

I have images with certain lines or bands on a ~white background. First I want to analyse the image to define the area of these lines or bands and then I would like to perform an analysis of a certain color range on that defined area only or those defined areas only.

What might be a good strategy for this?

Thanks for any help.

A

Please do not add topics to the main Digital Image Processing post at the top of this forum. I have separate your post. Also please do not post the same question to multiple forums. I have deleted your post in Fred's Scripts forum, since your question has nothing to do with those scripts.

What is your ImageMagick version and platform?

Please post and example image so we can understand better your question. You can upload to some free hosting service and put the URL here.


See the top-most post in the User's forum "IMPORTANT: Please Read This FIRST Before Posting" at viewtopic.php?f=1&t=9620

For novices, see

viewtopic.php?f=1&t=9620
http://http://www.imagemagick.org/scrip ... essing.php
http://www.imagemagick.org/Usage/reference.html
http://www.imagemagick.org/Usage/

As Fred says, we need to understand the problem before we can offer solutions.

For example, if the required area is a rectangle, then crop to it and use the in-built histograms.

My own process modules can regard alpha, so you can make pixels transparent to ignore them for counting, or make some 25% to count as 1/4 pixels, etc, like this:

(My apologies fmw42 for my ineptitude in posting to this forum. The post to Fred's Scripts was unintended and I could not quickly find the delete button. I hope that I have done better with this posting).

My problem is to achieve with an automated script, the location and the bounding of specific areas of interest in an image like www.fruitcom.com/imagemagick_stripes.tiff Here you see two darker bands and these are what I would like to "extract" for histogram and HSV and other statistical analysis to quantify a specific colour intensity.

The locations of these lines may vary in the image on the vertical axis, therefore the non-interactive script must not be dependent on a predetermined location. (It must define that itself and pass this definition to other processes).

How could this best be achieved with ImageMagick (to eventually port to PerlMagick)?

Version: ImageMagick 6.8.9-9 Q16 i686 2017-07-31
on Ubuntu 16.04.3 LTS
(libimage-magick-perl Version: 8:6.8.9.9-7ubuntu5)

Thanks for your support,
A

PS. Thanks for the resource links, many of the items in the left column of the root of this page have a server error, viz.
http://www.imagemagick.org/Usage/option ... gi?segment
And 404 here: http://http//www.imagemagick.org/script ... essing.php

Sorry for the typo. The correct link is http://www.imagemagick.org/script/comma ... essing.php

Here is one way.

Smooth out the image by averaging to one column, then expand horizontally.
Get the Chroma (saturation) channel, stretch to full dynamic range and threshold to make a black and white mask.
Use connected components analysis to extract the regions from the bounding boxes of the white areas in the mask from the original image.


tmp1 -- averaged image:


tmp2 -- chroma image:


tmp3 -- stretched image:


tmp4 -- thresholded image:


region 0:


region 1:



See http://magick.imagemagick.org/script/co ... onents.php

Dear Fred

Pure magick! And just what I ordered.
FWIW, on my platform this statement was never true:
`if [ "$color" = "gray(255)" ]; then'
because for example:
${arr[0]} = 0: 123x88+0+0 +61.0,+43.5 10824

But no matter, I removed it because I want to generate images of the different pieces.

Thank you for your assistance.

And thanks too, snigbo for your mkhisto.

Kind regards,
A

You are probably getting black and white rather than gray(0) and gray(255) or perhaps gray(0%) and gray(100%) or srgb(0,0,0) and srgb(255,255,255). You can check by doing
You may have to make tmp4.png bilevel.

If you are on IM 7, then change convert to magick.

Thank Fred.
It does not give me the colors in those formats with either of this options.
For the practical application at hand this is not a problem, as I will know what is a stripe and what is a background. And actually it is an advantage to have both the types of segments for analysis. In fact I want all the segments form the original, I just need to have the boundary information.

In fact, it occurs to me that I do not need to generate a physical image of each segment to be analysed, rather as long as I know the bounding box of that segment (from your mask), then I can use this information to perform all the analysis required on that segment of the original image. This is more direct and saves writing temporary files. I think.

FWIW, here are my outputs of your suggestions:
~ % convert --version
Version: ImageMagick 6.8.9-9 Q16 i686 2017-07-31 http://www.imagemagick.org
Copyright: Copyright (C) 1999-2014 ImageMagick Studio LLC
Features: DPC Modules OpenMP
Delegates: bzlib cairo djvu fftw fontconfig freetype jbig jng jpeg lcms lqr ltdl lzma openexr pangocairo png rsvg tiff wmf x xml zlib

~ % convert tmp4.png -type bilevel \
~ > -define connected-components:verbose=true \
~ > -define connected-components:mean-color=true \
~ > -connected-components 4 \
~ > null:
Objects (id: bounding-box centroid area):
1: 123x362+0+22 +61.0,+202.5 44526
0: 123x22+0+0 +61.0,+10.5 2706
~ % convert tmp4.png \
-define connected-components:verbose=true \
-define connected-components:mean-color=true \
-connected-components 4 \
null:
Objects (id: bounding-box centroid area):
1: 123x362+0+22 +61.0,+202.5 44526
0: 123x22+0+0 +61.0,+10.5 2706

I suspect your version of IM was just at the time that connected components was being developed and had not been fully debugged. I suggest that you upgrade if you can. Or perhaps if Linux, it has not been properly patched. I see you have 6.8.9.9, but the patch date says 2017-07-31. That means it has been patched (though not to the current level/date). But I suspect it has botched patches for connected components. 6.8.9.9 per se is about 120 versions old.

The first official debugged release was 6.8.9.10.

Thank you Fred, when I rolled my own ImageMagick 7.0.7-8 Q16 i686 2017-10-31 all now works as advertised.

A

fmw42 wrote: ↑2017-10-30T16:42:32-07:00 I see you have 6.8.9.9, but the patch date says 2017-07-31. That means it has been patched (though not to the current level/date). But I suspect it has botched patches for connected components. 6.8.9.9 per se is about 120 versions old.

The first official debugged release was 6.8.9.10.

I will take a wild guess and say the user is using ubuntu 16.04 (long term supported release), the version of IM is indeed 6.8.9.9 and compiled on 2017-07-31, but I think the only patches are security changes to files such as policy.xml.

The 6.8.9.9 I have ignores and the output looks this (note different image):
instead of

-define connected-components:mean-color=true was only added later at 6.9.2.8

See
http://magick.imagemagick.org/script/co ... onents.php
http://www.imagemagick.org/script/comma ... php#define (sections on connected components)