How to add a boundary into a image for saving

As the title, I have an image A and boundary B = bwboundaries(C) if I wanna show it, I can use imshow (A); hold on; visboundaries(B,'Color',''); if I use save button. the image I save don't have the same resolution as my source image so what should I do now?

Answer 1

[Gopichandh Danala]

The problem may just be with either the way you save or with way you used bwboundaries. I wrote some sample code using bwperim to get boundary image of a region then save it in png format to save lossless

img = imread('peppers.png');
figure, imshow(img);

% getting a segment 
h=imfreehand;
position = wait(h);
region=uint8(roipoly(img,position(:,1),position(:,2)));
region=region.*img;
figure,imshow(region); title('Segmented Region');

% use bwperim rather than bwboundaries to get an Image of boundary.
boundImg = uint8(bwperim(rgb2gray(region),4));
boundImg(boundImg ~=0) = max(img(:));
figure, imshow(img + boundImg)

% save image without loosing information..
imwrite(img + boundImg, 'writeImg.png');

% check 
read = imread('writeImg.png');
figure, imshow(read)
isequal(read, img + boundImg)

[R G]

it have an error on line "region=region.*img;"

[Gopichandh Danala]

Check the data type of both the variables region and img maybe they are different in your case and adjust

[R G]

I have got it, using:

region = cat(3,region,region,region); region=region.*img;

but the boundary line is too small, do you know how to make it bigger for making it clearer.

[Gopichandh Danala]

Just use imdilate on the boundary image.

% use bwperim rather than bwboundaries to get an Image of boundary.
boundImg = uint8(bwperim(rgb2gray(region),4));
boundImg = imdilate(boundImg, strel('disk',2)); % change strel size 2,3,.. as you need.. It will increase the boundary width..
boundImg(boundImg ~=0) = max(img(:));
figure, imshow(img + boundImg)

If you are satisfied with the answer please accept the answer

[R G]

perfect, thanks so much

[Gopichandh Danala]

no problem

[Warid Islam]

Hi Gopichandh,

Is it possible to crop the image specified by the boundary above?

[Image Analyst]

Of course. Use find() on region or position, then use max() and min() to find the max and min rows and columns, then use indexing to extract the cropped region.

[rows, columns] = find(region);
row1 = min(rows);
row2 = max(rows);
col1 = min(columns);
col2 = max(columns);
croppedImage = img(row1:row2, col1:col2, :);