Applying a circular mask to an image

I would like to take an image (attached original image) and apply a gray circular mask that has a soft fade-out (see reference image). I am quite new to working with natural images. Any insight / starter code would be much appreciated!

Answer 1

[DGM]

There are other posts about how to composite two images (or an image and a flat BG)

https://www.mathworks.com/matlabcentral/answers/890417-i-want-to-remove-black-color-from-my-image-and-make-the-background-transparent-so-that-i-can-paste-i

I'm going to do this one using MIMT tools for my own convenience.

% load foreground image
FG = imread('cameraman.tif'); % raw single-channel source

% make colored checkerboard background
sout = size(FG);
gradpict = lingrad([imsize(FG,2) 3],[0 1; 1 0],[0 0 1; 1 0 0]*255,'cosine','uint8'); % colored gradient
BG = im2uint8(0.3 + freecb(imsize(FG,2),[10 10])*0.4);
BG = imblend(gradpict,BG,1,'overlay');

% create soft mask
mask = radgrad(imsize(FG,2),[0.5 0.5],0.5,[255; 0]); % linear radial gradient
mask = imlnc(mask,'independent','in',[0 0.5],'k',2); % levels and curves tool

outpict = replacepixels(FG,BG,mask); % apply mask
imshow(outpict);

Adjusting the contrast ('k') parameter for imlnc() will increase the hardness of the mask. This is for k = 6:

In the above cases, I reused the colored checkerboard BG from the other example. Let's say you want a solid color BG instead:

% inputs to replacepixels() may be either images or tuples
%BG = colorpict(imsize(FG,2),[100 10 60],'uint8'); % an image
BG = [0.4 0.04 0.25]; % a tuple

Of course, simple gray backgrounds are easy enough to make without tools.

Most of the tools used above are from MIMT, which is here:

https://www.mathworks.com/matlabcentral/fileexchange/53786-image-manipulation-toolbox

If you need to know how they do what they do, just open up the files. In general, the main advantage to using things like replacepixels() is that it handles mismatches of class and number of channels, whereas naive multiplicative masking won't. Note that both these examples involve compositing a single-channel image with an RGB image using a single-channel mask.

[Blake Mitchell]

Thanks, this is super helpful! I was able to create exactly what I needed.