Is there some full code for the problem about pixel density for feature extraction?

Hay all, i have some 64x64 pixels image(bitscale) in different folder, try there some folder A,B,...until E. I want to extraction that image to get their feature and loop that extraction in every folder. Let's say 1 for black and 0 for white.

Here example of that image:

Deeply in my heart, i appreciate every help that you gave. Thank you very much.

Answer 1

[Image Analyst]

You can get a binary image like

binaryImage = grayImage < 128; % or whatever value works.

beyond that you can use regionprops() or somehow otherwise obtain the features you want, like morphology or whatever.

[Elder Winter]

Sir, i want to get value of that binary images like

0 0 0 0 0 0 0 0 0 9 0

0 0 0 0 0 0 0 0 6 0 3

0 0 0 0 0 0 0 1 0 0 4

and just use this code :

Image = imread('58a9cea0bee57.png');

we can get Image's value in workspace and when open it, there are val(:,:,1),val(:,:,2), and val(:,:,3). That val have same values. How to get just val(:,:,1) and copy to excel or another file format? Thank you :).

[Image Analyst]

That is a gray scale image, not a binary image. Binary images are of type logical and have values of only true or false (1 or 0).

Do not use Image as the name of your image. It's too much like image which is the name of a built in function.

If you have val, then you must have done

val = imread(filename);

To copy the red channel to an Excel workbook, use xlswrite()

xlswrite(filename, val(:,:,1));

[Elder Winter]

Ok, let's say that i have get my code like this.

binaryImage = imbinarize(grayImage);
xlswrite(vpixel,binaryImage);

How to automation this code to multifolder in one parent folder?

[Image Analyst]

See the FAQ: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

Or see my attached demo to recurse into subfolders.

[Elder Winter]

Then i get a task, that image 64x64 pixels must divided by 16 area. So every area has height 16 pixels and weight 16 pixels. I give it name Area I, II, ..., XV, XVI. So the question is typing 256 digits in every line then write it in txt and show up in matlab, then count how many number 1 in every area?

vpixel = fopen('pixelvalue.txt','w'); % 
formatSpec = '%d %d %d ';
imageTest = imread('a1.jpg');
grayImage = rgb2gray(imageTest);
binaryImage = imbinarize(grayImage);
reverseLogical = logical(1-binaryImage);
setInteger = double(reverseLogical);
[nrows,ncols] = size(setInteger);
for x=1:4
  for y=1:4
      for i=((nrows/4)*x-1)+1:((nrows/4)*(x))
          for j=((ncols/4)*y-1)+1:((ncols/4)*(y))               
              fprintf(vpixel,formatSpec,setInteger(i,j));
          end
          fprintf(vpixel,formatSpec,'\n');
      end
  end
end
type('pixelvalue.txt')

I have try to make the automation but fail. Please..

[Image Analyst]

Elder, I don't understand your question. If you have an image and want to count the 1's in it, you can do this:

numOnes = sum(yourArray(:)==1);

If you want to have that in a block that slides over your larger image without overlapping, then you can use blockproc, like in the attached examples.

[Elder Winter]

Ok, at least you a little know my question.

numOnes = sum(yourArray(:)==1);

Is that code can counting the 1's in function that blockproc? And what is 'yourArray' that for?

[Image Analyst]

That code counts 1's in the entire image. blockproc() counts the 1's in a window (part of the image). If you take the count from each window and make a 2-D array of them, you'll get another image where the pixel value is the count.

yourArray is whatever variable you have your image in. I don't know what you called it. I usually call mine "binaryImage" or "mask" or "grayImage" or "rgbImage" but you probably chose some other name for your variable. Replace yourImage with whatever the variable name is that you chose for your image.

[Elder Winter]

How about this code

counterFirst = sum(blockSize(:)==1);
blockImage = blockproc(setInteger,blockSize,counterFirst);

I don't know what kind of function is needed

[Elder Winter]

Ok, thank you so much.