Blur edges of rectangles in the image

I have the intensity image as shown bleow. I want to blur the corners of two rectangles (shown in image) so that it doesn't look like perfect rectangle.

How can i do that?

Answer 1

[Image Analyst]

kumara, try this. Adapt parameters as needed to control the size and amount of the blur.

% Demo to blur edges of rectangles, but not the insides.
clc;    % Clear the command window.
fprintf('Beginning to run %s.m ...\n', mfilename);
close all;  % Close all figures (except those of imtool.)
imtool close all;  % Close all imtool figures.
clear;  % Erase all existing variables.
workspace;  % Make sure the workspace panel is showing
fontSize = 15;

% Get image.
grayImage = abs(randn(224,224)+50);
m =2*abs(randn(90-10+1,60-10+1)+22); %first rectangle 
grayImage(10:90, 10:60) = m; 
m1 =3*abs(randn(110-95+1,60-10+1)+19); %second rectangle
grayImage(95:110, 10:60) = m1; 

[rows, columns, numberOfColorChannels] = size(grayImage)
subplot(2, 2, 1);
imagesc(grayImage)
impixelinfo; % Let user see RGB values as they mouse around.
colorbar
axis('on', 'image');
title('Original Image', 'FontSize', fontSize);
% Maximize the window to make it easier to draw.
g = gcf;
g.WindowState = 'maximized'
fprintf('Done running %s.m ...\n', mfilename);

% Get mask
mask = grayImage > 53 | grayImage < 47;
mask = imfill(mask, 'holes'); % Fill holes.
mask = bwareafilt(mask, 2); % Take 2 largest blobs only.
mask = bwconvhull(mask, 'objects'); % Take convex hull.
subplot(2, 2, 2);
% histogram(originalImage)
imshow(mask);
axis('on', 'image');
title('Mask Image', 'FontSize', fontSize);

% Get larger and smaller masks by dilating and eroding.
r = 3; % Radius for blur
se = strel('disk', r, 0);
bigMask = imdilate(mask, se);
smallMask = imerode(mask, se);
% xor to get outline.
borderMask = xor(bigMask, smallMask);
subplot(2, 2, 3);
imshow(borderMask);
axis('on', 'image');
title('Border Mask Image', 'FontSize', fontSize);

% Blur the entire image
windowSize = 5;
kernel = ones(windowSize) / windowSize^2;
blurredImage = imfilter(grayImage, kernel);
% Now replace the gray image inside the border mask with the blurred values.
grayImage(borderMask) = blurredImage(borderMask);
subplot(2, 2, 4);
imagesc(grayImage)
impixelinfo; % Let user see RGB values as they mouse around.
colorbar
axis('on', 'image');
title('Final Blurred Image', 'FontSize', fontSize);

[Jhe Mag]

Hi @Image Analyst what am I going to do if I am to input RGB image? I tried your code and run it with my image. It says errors like

Error using bwpropfilt

Expected input number 1, BW, to be two-dimensional.

[Image Analyst]

@Jhe Mag You can use imsplit() and then process each color channel one at a time.

[redChannel, greenChannel, blueChannel] = imsplit(rgbImage);

If you have any more questions, then start a new discussion thread and 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

[Raunak Gupta]

Hi,

If you know the corner points of rectangles in the image, then you can extract a matrix around those corners of rectangle with window size depending upon how much area you want to blur. Then you can apply imgaussfilt3 filter on that matrix which will make it blurred. Finally, you can replace the small matrix in the original image with the blurred matrix.

For example, in the image (size 224 x 224 x 3) given in the question one rectangle corner lies at (60,90) then let's say we take a window of 3 around that point.

I = imread('example_iamge.png'); 
windowSize = 3; 
sigma = 2; 
cornerPointMatrix = I(60-windowSize:60+windowSize,90-windowSize:90+windowSize,:); 
cornerPointMatrixBlur = imgaussfilt3(cornerPointMatrix, sigma); 
I(60-windowSize:60+windowSize,90-windowSize:90+windowSize,:) = cornerPointMatrixBlur; 

[kumara dommeti]

The images are intensity images (not RGB images). So i think 3D gaussian filter is not suitable. But imgaussfilt is useful. Here is my code for generating above image.

clc;
clearvars;
I = abs(randn(224,224)+50);
m =2*abs(randn(90-10+1,60-10+1)+22); %first rectangle 
I(10:90, 10:60) = m; 
m1 =3*abs(randn(110-95+1,60-10+1)+19); %second rectangle
I(95:110, 10:60) = m1; 

figure
imagesc(I)
colorbar

Now can you tell me how to use it?

[Raunak Gupta]

Hi,

You may look at Image Analyst solution, its detailed and very nice.