changing values of pixels in an image pixel by pixel ( Thresholding )

Hi everyone, I'm trying to put a threshold on a grayscale image, and I'm doing it this way: that a "for" loop reads the image pixel by pixel, and if the value of the pixel is less than "0.5" sets its' value to "0", and if it is more than "3", sets it to "256". Here it is my code:

my_image =imread('picture.tif');
for R=1:num of Rows
    for C=1:num of Columns
        pixel=my_image(R,C);
        if pixel<0.50000000 , pixel=0.000000000;, end
        if pixel>3.00000000 , pixel=256;, end
        thresh(R,C)=pixel;
    end
end
im_thresh=mat2gray(thresh);
figure,imshow(im_thresh);
title('thresholding');

But it doesn't work properly. Sometimes it misses some values that ought to be changed, and sometimes changes the ones that shouldn't be changed, to arbitrary values…for example changes the value "0.8182" to "0.0032". It also doesn't have enough accuracy, for example instead of changing the value "0.425" to "0", changes it to "0.0002". Could you please tell me what's causing this problem, and help me to fix the code?! Thanks in advance…

[Parth Batra]

I also need help with a similar probelem but i want to change the values of pixels between 135 and 165 to 0 please help

i tried the looping method but it's not providing satisfactory results.

[Waleed Naeem]

Please let me know also if you find the solution to your problem

[Image Analyst]

Waleed, since Mithra asked this 3 years ago, you might want to consider asking your own new question, and attach your image.

[Pravin Ninawe]

thanks

Answer 1

[Geert]

First of all, it is not necessary to loop over all pixels individually. You can simply use operations on your image matrix, which would look like this:

% read in tiff image and convert it to double format
my_image = im2double(imread('picture.tif'));
my_image = my_image(:,:,1);

% perform thresholding by logical indexing
image_thresholded = my_image;
image_thresholded(my_image>3) = 256;
image_thresholded(my_image<0.5) = 0;

% display result
figure()
subplot(1,2,1)
imshow(my_image,[])
title('original image')
subplot(1,2,2)
imshow(image_thresholded,[])
title('thresholded image')

The line image_thresholded(my_image>3) = 256; is an example of logical indexing, it will assign the value 256 to all matrix elements in image_thresholded for which the same matrix element in my_image is larger than 3.

If you prefer looping over all pixels (which is not advisable due to its slow speed), your code should look something like this:

% read in tiff image and convert it to double format
my_image = im2double(imread('picture.tif'));
my_image = my_image(:,:,1);

% allocate space for thresholded image
image_thresholded = zeros(size(my_image));

% loop over all rows and columns
for ii=1:size(my_image,1)
    for jj=1:size(my_image,2)
        % get pixel value
        pixel=my_image(ii,jj);

          % check pixel value and assign new value
          if pixel<0.5
              new_pixel=0;
          elseif pixel>3
              new_pixel=256;
          else
              new_pixel = pixel;
          end

          % save new pixel value in thresholded image
          image_thresholded(ii,jj)=new_pixel;
      end
  end

% display result
figure()
subplot(1,2,1)
imshow(my_image,[])
title('original image')
subplot(1,2,2)
imshow(image_thresholded,[])
title('thresholded image')

Does this help you out?

[Coco Coco]

And how can you do it with an RGB image? For a red, green and blue threshold

[Image Analyst]

Use the Color Thresholder app on the Apps tab of the tool ribbon.

[Mebiame]

Hello,

I was wondering if there was a way to use the above but for 3D CT images (images with slices) and only change the colour of a certain pixel on that certain slice.

Thanks

[Pravin Ninawe]

thanks

[Erin Browne]

Can I ask, if you are assigning a new pixel a number miy turns up red in colour, how do you change the color if you have multiple if requirements?

I.e. i want to change the colour of pixels to green, blue and red for three different pixel requirements.

its very similar to code above were you say new_pixel and its a value between 0-1.

[Image Analyst]

Can you start a new question, and attach your image, and your threshold range(s)? You'll need to create a new RGB based image based on your gray level image, and then add the colors based on your criteria.

[sowjanya]

@Image Analyst

like you said above i have converted from grayscale image to rgb image using ind2rgb

how to specify\assign colours to new pixels using thresholding method

[Image Analyst]

To give colors to a gray scale image, you'd have to convert it back to an RGB color image, either with cat() or ind2rgb(). With ind2rgb() you can supply a list of colors to assign every gray level to.

[Margi Sasono]

Very amazing. Thank you for your MATLAB code. I have one question. How to change value of pixels become a regular form, such as rectangular, triangular, cone, and others in an image?

[Imran Riaz]

@Image Analyst

I need your help in coding.

I have the finger knuckle images, I applied the mask function and remove the background. Now my supervisor asked me to remap the image with only those pixel values which have information and discard the zero valued pixels, mean background removal area is shown as zero pixels. How to do that? I share here images also.

[Image Analyst]

@Imran Riaz there are several ways you can do that. You can use regionprops and imcrop, or you can use any() or all() or sum() to find all-zero columns. For example:

mask = grayImage > 0;
props = regionprops(mask, 'BoundingBox');
grayImage = imcrop(grayImage, props.BoundingBox);

or

zeroColumns = all(grayImage == 0, 1);
grayImage(:, zeroColumns) = [];

[Imran Riaz]

This is my final masked image but the task is to remove the black area and the output contains 95% of finger pixels only. Bcz some pixels of black area will remain there.

[Image Analyst]

@Imran Riaz let's not continue to discuss here on @Mithra's thread (which sends him emails). Please start a new discussion thread and attach your original image. And say why you want it cropped and can't process it as it is (like you just want to magnify the ROI or something).

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

Answer 2

[Laurent]

Hi,

First I don't understand how your original image can have non-integer pixel values, because it seems your image is 8-bit from how you access the pixel-values. Is the value '0.8182' a pixel-value in your original image? Without the image itself it will be difficult to exactly determine what is going on.

Anyway, you can remove the for-loops and do it as follows:

thresh=my_image;
thresh(thresh<0.5)=0;
thresh(thresh>3)=255;

If your image is 8-bit, the values go from 0 to 255, not to 256.

[Image Analyst]

Her image is floating point (supposedly, but I'm not convinced) so if she's right then 256 is possible.

[Mithra]

actually it's a double image...

[Image Analyst]

OK. Floating point images (either single or double) can have a value of 256, whereas uint8 are limited to 255. But I see no reason at all why you would want that particular value. Can you explain why?

Answer 3

[Image Analyst]

Are you sure your tiff image has fractional values? I know it's a very flexible format and may allow floating point pixel values but it's not common. Anyway, you probably don't want a thresholded image with values of 0 and 256 and intact/unchanged values in the range 0.5 to 3.0. What are you going to do with the result anyway? Once I know that I can tell you how to proceed. You may want a binary image (the usual result of thresholding) to do what you need to do rather than another floating point image.

[Mithra]

I'm doing some morphological operations on this image, and in this point it's a double image. I want all pixel values less than 0.5 to be set on 0, and all pixel values more than 3 to be set as the highest value that a double image can have... which I assume is:

    realmax('double')

ans =

    1.7977e+308

the number 256 set by me previously was just an example indeed.

[Image Analyst]

But why? Are you familiar with morphology? Do you know how grayscale morphology differs from binary morphology (my guess is no)? OK, let me ask...after you do this grayscale morphology that you plan on doing, what do you want the result to be? Please upload an image and tell me what you want to find/measure/enhance in it.