How to crop a masked area (polygon shape) out of an image?

60 views (last 30 days)
Hi all,
Now my code is like below:
I = imread('moon.tif');
figure;
imshow(I)
Poly_Mask = images.roi.Polygon(gca,'Position',[100 150; 200 250; 300 350; 150 450]);
As you can see, the example image is now masked with a polygon shape. I would like to crop the masked area (polygon shape) out of the image. Could you please suggest me how to do this? Thank you.

Accepted Answer

Image Analyst
Image Analyst on 30 Mar 2022
Try this:
grayImage = imread('moon.tif');
[rows, columns, numberOfColorChannels] = size(grayImage)
rows = 537
columns = 358
numberOfColorChannels = 1
subplot(2, 1, 1);
imshow(grayImage, []);
axis('on', 'image');
title('Original Image')
x = [100, 200, 300, 150];
y = [150, 250, 350, 450];
% tack on starting point to close it so the plot is not oepn jaw. (Only
% needed if you wan to plot the outline in the overlay.
x = [x, x(1)];
y = [y, y(1)];
hold on;
plot(x, y, 'r-', 'LineWidth', 2);
mask = poly2mask(x, y, rows, columns);
% Erase outside mask by setting to zero
grayImage(~mask) = 0;
% Crop
croppedImage = grayImage(min(y):max(y), min(x):max(x));
subplot(2, 1, 2);
imshow(croppedImage)
axis('on', 'image');
title('Cropped Image')
  2 Comments
Image Analyst
Image Analyst on 30 Mar 2022
You can't make it transparent, and it does not need to be for any reason that I can think of, certainly not for normxcorr2().
You can have any shape you want just by putting in the coordinates of the shape boundaries into poly2mask().

Sign in to comment.

More Answers (1)

Tala
Tala on 29 Mar 2022
Edited: Tala on 29 Mar 2022
I cannot see your image. But I used a photo of my dog (because she wont sue me :D) to show the concept:
I=rgb2gray(imread("Tala1.jpg"));
x=[100 200 300 150];% your ROI x values
y=[150 250 350 450];% your ROI y values
[rows, columns,~] = size(I);
mask = imcomplement(poly2mask(x, y, rows, columns));
mul = immultiply(I,mask);
imshow(mul)
  3 Comments
Chutiphon Moranon
Chutiphon Moranon on 30 Mar 2022
Thank you for your reply. I will use normxcorr2 as you suggested. However, I would like to use an input image that is not rectangular in shape.
I am trying to crop an input image, only the part area I want for more accurate correlation. Therefore the background area of the cropped image will be transparent as the image below. I am not sure whether this method will provide good result but I will try it. Please feel free to provide your suggestions.Thank you.

Sign in to comment.

Products


Release

R2021a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!