apply logical mask to rgb image

I tried doing this:

foreground_mask = uint8(new_label == 1); 

segmented_foreground = imdata_orig .* repmat(foreground_mask, [1,1,3]);

and got a messed up original image, instead of the properly applied mask.

imdata_orig

is uint8 (which is why I make the logical -> uint8 casting).

What is wrong here? What are other ways to do this?

[Walter Roberson]

In what respect is it messed up? Are some values left non-zero when they should not be, or some made zero when they should not be, or does the result come out completely different than the mask ? Perhaps you could post some samples?

http://www.mathworks.com/matlabcentral/answers/7924-where-can-i-upload-images-and-files-for-use-on-matlab-answers

Answer 1

[Image Analyst]

Try Sean de Wolski's method:

% Mask the image.
maskedRgbImage = bsxfun(@times, rgbImage, cast(mask,class(rgbImage)));

where mask is a map of foreground and background like a logical image or a double image, that perhaps you got from color classification or something.

[Qazi Arbab Ahmed]

It does work for me last day. But now it displays "Undefined function or variable 'mask'.

Error in test_framereading (line 27) maskedRgbImage = bsxfun(@times, im, cast(mask,class(im)));"

[Jan]

@Qazi: "mask" is the logical array and you have to define this matrix by your own.

[Jan]

@Image Analyst: I thought several times to create a C-Mex function for the fast application of a logical mask to a multi-dim array. But I always stopped this, because I've been convinced, that the Image Processing Toolbox must contain such a function already, because the task is so common.

[Qazi Arbab Ahmed]

Thank You Simon. It worked when I defined it as bw. my binary Image. Thanks Alot, Now I want to extract temperature values from my ROI in a thermal video, Would you have any Idea?

[Image Analyst]

@Qazi Arbab Ahmed you'd have to extract one frame at a time and then convert the RGB image (if it's been pseudocolored) into a temperature image. See attached demos. Adapt as needed.

Answer 2

[Sam]

I had a similar problem. Try segmenting the channels first and applying the mask to each channel.

something from a similar project:

    r = im1(:,:,1);
    g = im1(:,:,2);
    b = im1(:,:,3);

    r(testgray<threshold) = 0;
    g(testgray<threshold) = 0;
    b(testgray<threshold) = 0;

    im1(:,:,1) = r;
    im1(:,:,2) = g;
    im1(:,:,3) = b;

testgray was a n*n matrix with the same width and height of im1, containing distance information from a 3D reconstruction, any values in the background (below the threshold) are set to 0, similar to masking, allowing only foreground image data.

[massonib]

Here is how I used the code above to solve the problem, where "I" is the rgb image.

imageSize = size(I);
ci = [200, 200, 100];     % center and radius of circle ([c_row, c_col, r])
[xx,yy] = ndgrid((1:imageSize(1))-ci(1),(1:imageSize(2))-ci(2));
mask = uint8((xx.^2 + yy.^2)<ci(3)^2);
r = I(:,:,1);
g = I(:,:,2);
b = I(:,:,3);
r = r.*mask;
g = g.*mask;
b = b.*mask;
I(:,:,1) = r;
I(:,:,2) = g;
I(:,:,3) = b;
imshow(I)

[Image Analyst]

Note: This is for the special case of a circular mask only.

Answer 3

[Eman]

Create a mask that extract the head of the cameraman that's located at the following locations : (100,40) (136,40) (100,80) (136,80) using AND logical operations.

Note: you can use

[x,y]= ginput(4) 

to get the coordinate of the cameraman head using the mouse cursor.

[Image Analyst]

@Eman this does not look like an answer to @Antonio's original post. This looks like your homework problem. If you have any questions ask your instructor or read the link below to get started:

How do I get help on homework questions on MATLAB Answers? - MATLAB Answers - MATLAB Central

Obviously we can't give you the full solution because you're not allowed to turn in our code as your own. But you can look at my attached demo. If you have any questions, then let's not send @Antonio more emails about your question, let's have you start your own separate thread.