How do I determine the percentage of pixels of a certain colour range within an area in my image?

I am trying to determine the percentage of pixels of a certain colour range within an area in my image.

I have already turned the image to HSI to to then find a range of intensity in which the clouds would appear as. I now want to find out what percentage of cloud (range of intensity) is within a certain region i.e. around the sun.

This in order to then use this information in an if statement to determine if a white is present within a specified region of the image.

Any help would be appreciated.

Answer 1

[Matt J]

 intensity = HSIimage(:,:,3);

 map= (intensity>=lowerlimit & intensity<=upperlimit);

 percentage=sum(map(:))/numel(map)*100;

[Jayden Hasemann]

Thanks for the reply, I think this is close to what I am looking for but not quite. The output of that is an array of numbers however i am not sure what that is telling me. essentially what i am looking at doing is use the percentage in an if statement to determine if a white is present within a specified region of the image.

[Matt J]

Whoops, sorry

   percentage=sum(map(:))/numel(map)*100;

[Jayden Hasemann]

Perfect, Thanks Matt

[Jayden Hasemann]

I do have a question about the code however, is it possible to go more indepth with the pixel value? what im working with is the 3 values for example in my HSI image purple is something similar to [0.09 0.06 0.71]

would something like this work?

 intensity = HSIimage(1,2,3);

 map= (intensity>=[lowerlimit1 lowerlimit2 lowerlimit3] & intensity<=[upperlimit1 upperlimit 2 upperlimit3]);
 percentage=sum(map(:))/numel(map)*100;

[Matt J]

Something like this, maybe, if you have R2016b or higher

vals=reshape(HSI,[],3);

   map= (vals>=[lowerlimit1 lowerlimit2 lowerlimit3] & vals<=[upperlimit1 upperlimit2 upperlimit3]);

percentage=sum(map,1)/numel(map)*100;

[Jayden Hasemann]

This is close Matt, thanks for that. Are you aware of a way of making it that each limit is reliant on the previous. like if a range of colours is set like lowerlimit1 upperlimit1 and then lowerlimit2 upperlimit2 and finally lowerlimit3 upperlimit3

to then give a singular percentage

[Matt J]

map2= all(map,2);
percentage=sum(map2,1)/numel(map2)*100

where map is as before.

[Jayden Hasemann]

perfect Matt, thanks a heap

Answer 2

[Image Analyst]

I don't think your algorithm is very robust. I suggest you follow the papers of researchers who have successfully found clouds and published their algorithms. See articles on cloud detection in Vision Bib:

http://www.visionbib.com/bibliography/motion-f716.html#Tracking%20for%20Weather,%20Clouds

16.6.2.12 Tracking for Weather, Clouds

[Jayden Hasemann]

Thanks for that, I see you have a very in depth knowledge in DSP with matlab. do you have any examples of finding clusters of a certain range of intensities within an image?

[Image Analyst]

Try the colorcloud() function. Also try the classification Learner App on the Apps tab. Also try the Color Thresholder on the Apps tab.

And see my color segmentation demos on my File Exchange http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862&sort=downloads_desc