How can blur an image

433 views (last 30 days)
Jorge Ortiz
Jorge Ortiz on 16 Mar 2019
Edited: DGM on 28 Feb 2022
Write a function called blur that blurs the input image. The function is to be called like this:
output = blur(img,w);
where img, the input image is a two-dimensional matrix of grayscale pixel values between 0 and 255. Blurring is to be carried out by averaging the pixel values in the vicinity of every pixel. Specifically, the output pixel value is the mean of the pixels in a square submatrix of size 2w+1 where the given pixel sits in the center. For example, if w is 1, then we use a 3x3 matrix, that is, we average all the neighboring pixels of the given pixel and itself. Only use valid pixels when portions of the blurring matrix fall outside the image. For example, the blurred value corresponding to w = 1 at index (1,1) would be the mean of of elements (1,1), (1, 2), (2,1) and (2, 2). Both input img and output output are of type uint8.
You can download the test image here to use in MATLAB.
  23 Comments
Image Analyst
Image Analyst on 4 Jun 2021
You know, there is a little double page icon in the upper left of the code blocks so that you can copy the code. The code blocks I see here all have uint8(), not unit8(). If you had copied, you would not have had that problem.
Sana Hafeez
Sana Hafeez on 5 Jun 2021
Thank you so much

Sign in to comment.

Answers (13)

Dipesh Poudel
Dipesh Poudel on 9 Apr 2020
function [output] = blur(A,w)
[row col] = size(A);
A=uint8(A);
B=nan(size(A) + (2*w));
B(w+1:end-w,w+1:end-w)=A;
output = 0*A;
for i=w+1:row+w
for j=w+1:col+w
tmp=B(i-w:i+w,j-w:j+w);
output(i-w,j-w)=mean(tmp(~isnan(tmp)));
end
end
output=uint8(output);
  6 Comments
Sourabh Bhange
Sourabh Bhange on 10 Feb 2021
Edited: DGM on 28 Feb 2022
Hello, Thank you for the reply.
Using the function you suggested:
function output = blur(grayImage, w)
% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(grayImage);
% Make sure it's a gray scale image, not a color image.
if numberOfColorChannels > 1
% It's not really gray scale like we expected - it's color.
% Use weighted sum of ALL channels to create a gray scale image.
grayImage = rgb2gray(grayImage);
% ALTERNATE METHOD: Convert it to gray scale by taking only the green channel,
% which in a typical snapshot will be the least noisy channel.
% grayImage = grayImage(:, :, 2); % Take green channel.
end
B = nan(size(grayImage) + (2*w)); % Make B larger, padding with NaNs.
B(w+1:end-w, w+1:end-w) = grayImage; % Make central part of B match A.
output = double(grayImage); % Initialize a double array for the output with the size and values of the input.
% Scan the window over the image making the output the average of the input at each pixel location.
for row = w+1:rows+w
for col = w+1:columns+w
% Extract the part of the image under the window at the window's current location.
tmp = B(row-w:row+w, col-w:col+w);
% Replace the output's value there with the mean of the input.
output(row-w, col-w) = mean(tmp(:), 'omitnan');
end
end
% Cast output image to be the same class (uint8, uint16, single, or double) as the input image.
output = cast(output, 'like', grayImage);
When I used in my code I am able to convert rgb2gray and gray2blur.
How I can modify this function if I want to conver rgb2blur?
Image Analyst
Image Analyst on 11 Feb 2021
Your MyBlur() function, which takes an RGB image needs to call imsplit() and call your blur(), which deals with monochrome images:
function blurredRGBImage = MyBlur(rgbImage, windowWidth)
% Split into separate color channels.
[redChannel, greenChannel, blueChannel] = imsplit(rgbImage);
% Blur each color channel independently.
blurredR = blur(redChannel, windowWidth);
blurredG = blur(greenChannel, windowWidth);
blurredB = blur(blueChannel, windowWidth);
% Recombine into a single RGB image.
blurredRGBImage = cat(3, blurredR, blurredG, blurredB);
Then call it:
rgbImage = imread('peppers.png');
blurredRGBImage = uint8(MyBlur(rgbImage, 9));
imshow(blurredRGBImage)

Sign in to comment.


Yaksha SJ
Yaksha SJ on 10 May 2020
function out = blur(img,w)
% convert to double for doing calculations
imgD = double(img);
[row, col] = size(img);
out = zeros(row, col);
for ii = 1:row
for jj = 1:col
% Get the indices for a submatrix
r1 = ii-w;
r2 = ii+w;
c1 = jj-w;
c2 = jj+w;
% Test that indices are valid
% If not, set to min/max that is valid
if r1 < 1
r1 = 1;
end
if r2 > row
r2 = row;
end
if c1 < 1
c1 = 1;
end
if c2 > col
c2 = col;
end
% Get the submatrix and assign the mean to the output pixel
m = imgD(r1:r2, c1:c2);
out(ii,jj) = mean(m(:));
end
end
% convert back to uint8
out = uint8(out);
end
  2 Comments
Walter Roberson
Walter Roberson on 10 May 2020
hint:
r1 = max(1, ii-w);
Kartik agarwal
Kartik agarwal on 14 May 2020
Edited: Kartik agarwal on 14 May 2020
Hey, if I produce this change in the last segment of your code, I am getting errors :
Original code :
% Get the submatrix and assign the mean to the output pixel
m = imgD(r1:r2, c1:c2);
out(ii,jj) = mean(m(:));
end
end
% convert back to uint8
out = uint8(out);
end
Changed code :
% Get the submatrix and assign the mean to the output pixel
m = imgD(r1:r2, c1:c2);
out(ii,jj) = uint8(mean(m(:))); % uint8 conversion done here itself%
end
end
end
The error says that Variable output must be of data type uint8. It is currently of type double.
I am new to Matlab. I can't understand the error in doing conversion the way I did in changed code. Please help

Sign in to comment.


KSSV
KSSV on 16 Mar 2019
Edited: KSSV on 16 Mar 2019
I = imread('peppers.png') ;
for i = 1:3
I(:,:,i) = uint8(conv2(I(:,:,i),ones(20)/20^2,'same'));
end
imshow(I) ;
  4 Comments
Gloria Paul
Gloria Paul on 22 Jun 2020
Hello Jorge, I really need help figuring this code out
Image Analyst
Image Analyst on 23 Jun 2020
What code? Jorge's or KSSV's? Or one of the other multitude of places where code appears on this page?

Sign in to comment.


Image Analyst
Image Analyst on 16 Mar 2019
Can your function call the built-in functions conv2() or imfilter()?
If not, see, and adapt, my attached manual convolution program.
  2 Comments
Jorge Ortiz
Jorge Ortiz on 16 Mar 2019
Edited: Image Analyst on 4 May 2020
Already done!! Above in the question is the solution I chose.
Thank you very much anyway.
Image Analyst
Image Analyst on 22 Mar 2019
Edited: Image Analyst on 4 May 2020
Well the solution you chose is not very robust, but it will work in some cases, not all cases.

Sign in to comment.


Jaimin Motavar
Jaimin Motavar on 4 Jul 2019
can someone help me to figure out what is wrong in this code?
function output = blur(img,w)
B=double(img);
[m,n] = size(B);
k=2*w+1;
for i = 1:m
for j = 1:n
p=i-fix(k/2);
q=i+fix(k/2);
r=j-fix(k/2);
s=j+fix(k/2);
if p<1
p=1;
end
if q>m
q=m;
end
if r<1
r=1;
end
if s>n
s=n;
end
A=B([p:q],[r:s]);
B(i,j)=mean(A(:));
end
end
output=uint8(B);
end
  5 Comments
Namarta Kapil
Namarta Kapil on 21 Jul 2019
Can somebody please tell me what is wrong with this code? it gives the following error:
  • Assessment result: incorrectSimple testVariable output has an incorrect value. Tested with a 5x5 uint8 matrix with all 255-s in the first and last rows and columns and zeros everywhere else and w = 1
  • Assessment result: incorrectUsing image fileVariable output has an incorrect value. Tested with the vandy.png file and w = 1
Walter Roberson
Walter Roberson on 23 Jul 2019
You are overwriting your input data in B as you do the calculation, so your calculation is being based upon the already-smoothed pixels above and to the left instead of on the original pixels above and to the left.

Sign in to comment.


Brandon McLaughlin
Brandon McLaughlin on 22 Jul 2019
Help! I'm almost done with this course but for some reason I can't get this function to work. I have my code all set, but it keeps saying I'm doing it wrong. What am I doing wrong?
function output=blur(img,w)
img=double(img);
d=w*2+1
s=size(img)
output=[]
for r=1:d:s(1)
row=[];
for c=1:d:s(2)
if r+w<=s(1) && c+w<=s(2) && r-w>=1 && c-w>=1
tot=sum(sum(img((r-w):(r+w),(c-w):(c+w))));
val=tot/(d*d);
nxt=ones(d,d);
nxt(1:end,1:end)=val;
row=[row nxt];
elseif c+w<=s(2) && r-w>=1 && c-w>=1
tot=sum(sum(img((r-w):end,(c-w):(c+w))));
val=tot/(d*(s(1)-r+w));
nxt=ones((s(1)-r+w),d);
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r+w<=s(1) && r-w>=1 && c-w>=1
tot=sum(sum(img((r-w):(r+w),(c-w):end)));
val=tot/(d*(s(2)-c+w));
nxt=ones(d,(s(2)-c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r+w<=s(1) && c+w<=s(2) && r-w>=1
tot=sum(sum(img((r-w):(r+w),1:(c+w))));
val=tot/(d*(c+w));
nxt=ones(d,(c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r+w<=s(1) && c+w<=s(2) && c-w>=1
tot=sum(sum(img(1:(r+w),(c-w):(c+w))));
val=tot/(d*(r+w));
nxt=ones((r+w),d);
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r+w<=s(1) && c+w<=s(2)
tot=sum(sum(img(1:(r+w),1:(c+w))));
val=tot/((r+w)*(c+w));
nxt=ones((c+w),(r+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r-w>=1 && c-w>=1
tot=sum(sum(img((r-w):end,(c-w):end)));
val=tot/((s(1)-r+w)*(s(2)-c+w));
nxt=ones((s(1)-r+w),(s(2)-c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif r+w<=s(1) && c-w>=1
tot=sum(sum(img(1:(r+w),(c-w):end)));
val=tot/((r+w)*(s(2)-c+w));
nxt=ones((r+w),(s(2)-c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif c+w<=s(2) && r-w>=1
tot=sum(sum(img((r-w):end,1:(c+w))));
val=tot/((s(1)-r+w)*(c+w));
nxt=ones((s(1)-r+w),(c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
end
end
output=[output;row];
end
row=[]
s2=size(output)
if s2(1)<s(1)
for c=1:d:s(2)
if c+w<=s(2) && c-w>=1
tot=sum(sum(img((s2(1)+1):s(1),(c-w):(c+w))));
val=tot/((s(1)-s2(1))*d);
nxt=ones((s(1)-s2(1)),d);
nxt(1:end,1:end)=val;
row=[row nxt];
elseif c+w<=s(2)
tot=sum(sum(img((s2(1)+1):s(1),1:(c+w))));
val=tot/((s(1)-s2(1))*(c+w));
nxt=ones((s(1)-s2(1)),(c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
elseif c-w>=1
tot=sum(sum(img((s2(1)+1):s(1),(c-w):end)));
val=tot/((s(1)-s2(1))*(s2(2)-c+w));
nxt=ones((s(1)-s2(1)),(s2(2)-c+w));
nxt(1:end,1:end)=val;
row=[row nxt];
end
end
output=[output;row];
end
row=[]
col=[]
s3=size(output)
if s3(2)<s(2)
for r=1:d:s(1)
if r+w<=s(1) && r-w>=1
tot=sum(sum(img((r-w):(r+w),(s3(2)+1):s(2))));
val=tot/(d*(s(2)-s3(2)));
nxt=ones(d,(s(2)-s3(2)));
nxt(1:end,1:end)=val;
col=[col;nxt];
elseif r+w<=s(1)
tot=sum(sum(img(1:(r+w),(s3(2)+1):s(2))));
val=tot/((r+w)*(s(2)-s3(2)));
nxt=ones((r+w),(s(2)-s3(2)));
nxt(1:end,1:end)=val;
col=[col;nxt];
elseif r-w>=1
tot=sum(sum(img((r-w):(r+w),(s3(2)+1):s(2))));
val=tot/((s3(1)-r+w)*(s(2)-s3(2)));
nxt=ones((s3(1)-r+w),(s(2)-s3(2)));
nxt(1:end,1:end)=val;
col=[col;nxt];
end
end
s4=size(col)
if s4(1)<s3(1) || s4(1)<s(1)
tot=sum(sum(img((s2(1)+1):s(1),(s3(2)+1):s(2))));
val=tot/((s(1)-s2(1))*(s(2)-s3(2)));
nxt=ones((s(1)-s2(1)),(s(2)-s3(2)));
nxt(1:end,1:end)=val;
col=[col;nxt];
end
output=[output col];
end
output=uint8(output);

Image Analyst
Image Analyst on 23 Jul 2019
See my manual convolution routine, attached.

shreyansh pitroda
shreyansh pitroda on 29 Apr 2020
%% can some one find what is error in this code
function [output] = blur(img,w)
img = uint8(img);
[a,b] = size(img);
output = zeros(a,b);
h = 2*w+1;
for i = 1:a-h
for j = 1:b-h
value = img(i:(h+i-1),j:(h+j-1));
output(i,j) = mean(value(:));
end
end
output = uint8(output);
end
  6 Comments
Image Analyst
Image Analyst on 4 Jun 2021
@Sana Hafeez, post your code, because I just copied and pasted my code above and it works great. You must have defined blur() differently. Is it in the same file as your script, or a separate m-file. What does this show:
>> which -all blur
Again, here is my code that works perfectly fine.
grayImage = imread('cameraman.tif');
hFig = figure;
subplot(1, 2, 1);
imshow(grayImage);
title('Original', 'FontSize', 20);
hFig.WindowState = 'maximized';
windowHalfWidth = 9;
output = blur(grayImage, windowHalfWidth);
subplot(1, 2, 2);
imshow(output);
title('Blurred', 'FontSize', 20);
% can some one find what is error in this code
function output = blur(img, w)
img = uint8(img);
[rows, columns] = size(img);
output = img; % Initialize output to be the same size as the input image.
h = 2 * w + 1;
fprintf('Filter is %d rows by %d columns.\n', h, h);
for col = h : columns-h
for row = h : rows-h
value = img(row : (h+row-1), col : (h+col-1));
output(row, col) = mean(value(:));
end
end
output = uint8(output);
end
Sana Hafeez
Sana Hafeez on 5 Jun 2021
Thanks a lot!!

Sign in to comment.


David Gonzalez
David Gonzalez on 25 May 2020
function output = blur(img,w)
mirror = -1*ones(size(img) + 2*w);
mirror(w+1:end-w,w+1:end-w) = img;
output = zeros(size(img));
for ii = w+1:size(mirror,1)-w
for jj = w+1:size(mirror,2)-w
submatrix = mirror(ii-w:ii+w,jj-w:jj+w);
sum_values = sum(submatrix(submatrix>=0));
valid = sum(submatrix(:)>=0);
output(ii-w,jj-w) = sum_values/valid;
end
end
output = uint8(output);

Divya Ratna
Divya Ratna on 25 May 2020
i figured it out by myself...
i know its too lengthy..
function out = blur (img, w)
temp = img;
s = size(img);
if s(1) > w && s(2) > w
%1
for i = 1 : w
for j = 1 : w
temp (i,j) = sum(sum(img(1:i+w , 1:j+w))') / ((i+w)*(j+w));
end
end
%2
for i = 1 : w
for j = w+1 : s(2)-w
temp (i,j) = sum(sum(img(1:i+w , j-w:j+w))') / ((i+w)*(w+w+1));
end
end
%3
for i = 1 : w
for j = s(2)-w+1 : s(2)
temp (i,j) = sum(sum(img(1:i+w , j-w:s(2)))') / ((i+w)*(s(2)-j+1+w));
end
end
%4
for i = w+1 : s(1)-w
for j = 1 : w
temp (i,j) = sum(sum(img(i-w:i+w , 1:j+w))') / ((w+w+1)*(j+w));
end
end
%5
for i = w+1 : s(1)-w
for j = w+1 : s(2)-w
temp (i,j) = sum(sum(img(i-w:i+w , j-w:j+w))') / ((w+w+1)*(w+w+1));
end
end
%6
for i = w+1 : s(1)-w
for j = s(2)-w+1 : s(2)
temp (i,j) = sum(sum(img(i-w:i+w , j-w:s(2)))') / ((w+w+1)*(s(2)-j+1+w));
end
end
%7
for i = s(1)-w+1 : s(1)
for j = 1 : w
temp (i,j) = sum(sum(img(i-w:s(1) , 1:j+w))') / ((s(1)-i+1+w)*(j+w));
end
end
%8
for i = s(1)-w+1 : s(1)
for j = w+1 : s(2)-w
temp (i,j) = sum(sum(img(i-w:s(1) , j-w:j+w))') / ((s(1)-i+1+w)*(w+w+1));
end
end
%9
for i = s(1)-w+1 : s(1)
for j = s(2)-w : s(2)
temp (i,j) = sum(sum(img(i-w:s(1) , j-w:s(2)))') / ((s(1)-i+1+w)*(s(2)-j+1+w));
end
end
end
out = temp;
end

Chandrashekhar Dhangar
Chandrashekhar Dhangar on 16 Jul 2020
%can someone pls help with this?
%Error=Variable output has an incorrect value.
%Tested with a 5x5 uint8 matrix with all 255-s in the first and last rows and columns and zeros everywhere else and w = 1
function output=blur(img,w)
[m, n]=size(img);
im=double(img);
S=zeros(2*w+1,2*w+1);
out=zeros(m,n);
for i=1:m
for j=1:n
for u=1:2*w+1
for v=1:2*w+1
if i-w>0 && j-w>0 && i-w+u-1<=m && j-w+v-1<=n
S(u,v)=im(i-w+u-1,j-w+v-1);
else
S(u,v)=0;
end
end
end
out(i,j)=sum(S(:))/(2*w+1)^2;
end
end
output=uint8(out);
end
  9 Comments

Sign in to comment.


aniket GIRI
aniket GIRI on 18 Jul 2020
Edited: DGM on 28 Feb 2022
function out = blur(img,w)
% convert to double for doing calculations
imgD = double(img);
[row, col] = size(img);
out = zeros(row, col);
for ii = 1:row
for jj = 1:col
% Get the indices for a submatrix
r1 = ii-w;
r2 = ii+w;
c1 = jj-w;
c2 = jj+w;
% Test that indices are valid % If not, set to min/max that is valid
if 1 < 1
r1 = 1;
end
if r2 > row
r2 = row;
end
if c1 < 1
c1 = 1;
end
if c2 > col
c2 = col;
end
% Get the submatrix and assign the mean to the output pixel
m = imgD(r1:r2, c1:c2);
out(ii,jj) = mean(m(:));
end
end
% convert back to uint8
out = uint8(out);
end
  4 Comments
Deepthi Thomas
Deepthi Thomas on 23 Jul 2020
I changed the mean variable to another name 'mean1'. I checked with the debugger and the loop is running till row=1 and col=3.
But it then stops with another error message:
Index in position 1 is invalid. Array indices must be positive integers or logical values.
Error in blur (line 22)
sub_matrix= img1(row-w:row+w,col-w:col+w);
The sub_matrix remains a 3X4 double and the mean1 is a 1X1 double, 255, since I changed mean1= mean(sub_matrix(:)). But the debugger goes to the else statement and shows this error message.
Image Analyst
Image Analyst on 23 Jul 2020
If row is 1, then row-w is zero or negative. You can't start at row 1 unless you check for the case where the window would be off the image, in which case you'd have to truncate the window. See my attached manual convolution.

Sign in to comment.


Josh Yaksich
Josh Yaksich on 13 Jun 2021
Edited: Josh Yaksich on 13 Jun 2021
This is what I did and it worked. Please don't copy it directly, as you will only be hurting yourself. Hopefully it helps when deciding what approach to take:
function output = blur(img,w)
%save dimensions of img
y = size(img,1);
x = size(img,2);
%make a duplicate img for modification
imgdup = img;
%create a larger matrix of all -1's that's the same size as image plus a border
bsquare = ones(y + 2 * w,x + 2 * w) * -1;
%copy img inside of larger matrix
bsquare((w + 1):end - w, w + 1:end - w) = img;
for i = 1:y
for j = 1:x
%make a small matrix for an individual pixel plus its surrounding pixels
pixelraw = bsquare([i:2 * w + i],[j:2 * w + j]);
%calculate the average value in the small matrix while ignoring -1
pixelavg = mean(pixelraw(pixelraw ~= -1));
%copy the average value to an individual pixel in the duplicate image
imgdup(i,j) = pixelavg;
end
end
%convert duplicate image to unint8 and assign to output
output = uint8(imgdup);

Community Treasure Hunt

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

Start Hunting!