Introduction to MATLAB programming - image blur problem
Show older comments
Hello all,
I am trying to work through one of the problems for the 'introduction to matlab programming' course by coursera, known as the 'image blur' problem (lesson 8 - final problems, week 9). The system keeps telling me that my values are incorrect, but I am struggling to understand why? I have attached my code below. I would be very grateful if anyone could give me a helping hand.
%The code to call the function:
img = imread('vandy.png')
output = blur(img,2) % I have specified the function 'blur' below
imshow(output);
% The function:
function[output] = blur(img,w)
here = 2*w + 1;
change = w;
%setting an epmpty matrix
emptymatrix = zeros(length(img(:,1)),length(img(1,:)));
%working out the output pixel value for each corner
added = img((1):(1+change),(1):(1+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(1,1) = answer;
i = length(img(:,1));
z = 1;
added = img((i-change):(i),(z):(z+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(i,z) = answer;
i = 1;
z = length(img(1,:));
added = img((i):(i+change),(z-change):(z));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(i,z) = answer;
i = length(img(:,1));
z = length(img(1,:));
added = img((i-change):(i),(z-change):(z));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(i,z) = answer;
output = uint8(emptymatrix);
% working out the pixel values for the inner square
for i = 1:length(img(:,1))
for z = 1:length(img(1,:))
if ((i + change) <= length(img(:,1))) && ((i-change) > 0) && ((z+change) <= length(img(1,:))) && ((z-change) > 0)
added = img((i-change):(i+change),(z-change):(z+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/(here*here);
answer = uint8(mean);
emptymatrix(i,z) = answer;
output = uint8(emptymatrix);
end
end
end
% working out the output pixel values for the outer rows and columns (excluding the corners)
for i = 2:(length(img(:,1))-1)
if (i-change) > 0 && (i+change) < length(img(:,1))
z = 1;
added = img((i-change):(i+change),(z):(z+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(i,z) = answer;
output = uint8(emptymatrix);
end
end
for i = 2:(length(img(:,1))-1)
if (i-change) > 0 && (i+change) < length(img(:,1))
z = length(img(1,:));
added = img((i-change):(i+change),(z-change):(z));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(i,z) = answer;
output = uint8(emptymatrix);
end
end
for i = 2:(length(img(1,:))-1)
if (i-change) > 0 && (i+change) < length(img(1,:))
z = 1;
added = img((z):(z+change),(i-change):(i+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(z,i) = answer;
output = uint8(emptymatrix);
end
end
for i = 2:(length(img(1,:))-1)
if (i-change) > 0 && (i+change) < length(img(1,:))
z = length(img(:,1));
added = img((z-change):(z),(i-change):(i+change));
addition = sum(added);
additionally = sum(addition);
mean = additionally/numel(added);
answer = uint8(mean);
emptymatrix(z,i) = answer;
output = uint8(emptymatrix);
end
end
end
5 Comments
Geoff Hayes
on 8 Apr 2020
ey21 - which values are considered to be incorrect? The output? The above is a lot of code to go through without knowing why it does what it does. I did try it with a jpeg and noticed that my input 260x396x3 image became a 260x1188 output image (where 1188=3*396). Are you assuming that your image inputs are 2D grayscale only? Is that true for the vandy.png that you are testing with? Or does the code need to be able to handle colour input images?
ey21
on 8 Apr 2020
Geoff Hayes
on 9 Apr 2020
Then there is probably something wrong with your algorithm. Can you add some comments to describe what the code is doing?
Image Analyst
on 9 Apr 2020
Can you use imfilter() or conv2()? It would be a lot shorter.
Accepted Answer
Geoff Hayes
on 9 Apr 2020
Edited: Geoff Hayes
on 9 Apr 2020
ey21 - I think that you need to simplify your code. For the block that is concerned with the inner rows and columns, I've reduced it to
for i = 1:length(img(:,1))
for z = 1:length(img(1,:))
if ((i + w) <= length(img(:,1))) && ((i-w) > 0) && ((z+w) <= length(img(1,:))) && ((z-w) > 0)
submatrix = img((i-w):(i+w),(z-w):(z+w));
output(z,i) = uint8(sum(submatrix(:))/numel(submatrix));
end
end
end
There is only one sum since submatris(:) will return a column of all elements in the submatrix, and then we can just sum those elements. numel is used to return the number of elements in the submatrix so that we can do the appropriate average. I've eliminated the emptymatrix since it is redundant. (I've replaced the change with w since they are identical.)
You still have two other blocks of code, so can all of this be simplified further? I think so...so long as you choose the correct submatrix for each element in the input matrix. In fact, I think that you can do all of this work in one loop regardless as to the pixel if you choose the correct minimum and maximum row and column to create the submatrix from. The example in your assignment tells you what to do if your code is considering the pixel at (1,1) so you could use that information to determine what to do.Your code to get the submatrix of an inner pixel is
submatrix = img((i-w):(i+w),(z-w):(z+w));
because you know that i-w is valid, i+w is valid, z-w is valid, and z+w is valid. This could be made more general if you do
minValidRowIndex = max(1,i-w);
maxValidRowIndex = min(numRows,i+w);
minValidColIndex = max(1,z-w);
maxValidColIndex = min(numCols,z-w);
The above four variables can now be used to get your submatrix for any (I think) pixel. We use max and min to ensure that we never use row or column indices that are outside of the matrix.
2 Comments
ey21
on 13 Apr 2020
Geoff Hayes
on 14 Apr 2020
Do you really need to consider all of those cases? I really don't think that you need any if/elseif statements and so could perhaps reduce the code to simply
for z = 1:length(img(1,:)) % iterating over columns
for i = 1:length(img(:,1)) % iterating over rows
minValidRowIndex = max(1,i-w);
maxValidRowIndex = min(numRows,i+w);
minValidColIndex = max(1,z-w);
maxValidColIndex = min(numCols,z-w);
submatrix = img(minValidRowIndex:maxValidRowIndex, minValidColIndex:maxValidColIndex);
output(z,i) = uint8(sum(submatrix(:))/numel(submatrix));
end
end
More Answers (1)
Swapnil Yadav
on 22 Jun 2020
2 votes
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
Aindrila Bose
on 23 Aug 2020
Thanks a lot. Code was easy to understand.
Summer K.Rock
on 24 Jan 2022
this answer is super easy and simple !! very good !!
Categories
Find more on Matrix Indexing in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
