change an image from rectangular to square by adding white area
显示 更早的评论
Dears, i have a dataset that im working on but all of them in rectangular shape, i wand them to be square by adding white area ( raws or coloumns depending on the image) many thanks in advanced
2 个评论
Rik
2022-11-17
What have you tried so far? This type of padding should not be too hard. Note that the value that you need to add may depend on the data type and value range of the image.
hamsa dhia
2022-11-17
编辑:hamsa dhia
2022-11-17
回答(2 个)
The documentation for padarray states it will pad with 0. If you don't want that, you will have to provide the value as the third argument. You probably want 255, but you can use the code below to determine that dynamically.
switch class(im)
case 'uint8'
padval = uint8(255);
case 'double'
padval = 1;
otherwise
error('check what value you need')
end
Edit: a much better solution that this code is to use the function that @DGM showed in his answer: getrangefromclass.
range = getrangefromclass(im)
padval = range(2);
3 个评论
hamsa dhia
2022-11-17
Rik
2022-11-17
What intensity value does the white part have in your image?
hamsa dhia
2022-11-17
I don't normally use getrangefromclass(), but that's one way to get the white value. Of course, it's impossible to tell that the result is padded here because it's displayed on a white background
inpict = imread('peppers.png');
%inpict = rot90(inpict); % will pad either dimension
outpict = squarewhitematting(inpict);
imshow(outpict)
function outpict = squarewhitematting(inpict)
% add symmetric white padding to the smaller dimension of the image geometry
% so that the resulting image is square
% INPICT is an image of any standard image class, with any number of channels.
% multiframe images are supported
% get nominal black and white value
valuerange = getrangefromclass(inpict);
[h,w,~] = size(inpict);
if h == w
% nothing to do
outpict = inpict;
return;
else
if w > h % wide image; pad top/bot
padamt = [(w-h)/2 0];
else % tall image; pad l/r
padamt = [0 (h-w)/2];
end
% use floor/ceil to deal with odd differences
outpict = padarray(inpict,floor(padamt),valuerange(2),'pre');
outpict = padarray(outpict,ceil(padamt),valuerange(2),'post');
end
end
4 个评论
If you use this code instead of imshow you can verify that there is a white border added.
image(outpict),daspect([1 1 1])
DGM
2022-11-17
That's a good point
hamsa dhia
2022-11-17
Rik
2022-11-17
Something strange is going on if that line doesn't work. image is the primitive that is called by imshow (just like plot calls line and surf calls patch).
But if you have a working solution now, that should do it. There is no fundamental difference between the code I suggested and what DGM implemented (other than that DGM provided you with a full implementation).
类别
在 帮助中心 和 File Exchange 中查找有关 Image Arithmetic 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!


