Alphamap using Binary Mask

3 views (last 30 days)
Sordin
Sordin on 22 Apr 2017
Commented: Walter Roberson on 24 Apr 2017
Is it possible to make an alphamap for a figure based on its binary mask?
Here is a sample image and its binary:
I = imread('saturn.png');
a = im2bw(I, graythresh(I));
I want the regions in black to be made transparent and white regions to be opaque. So the only area visible is the image itself without the background.
Is this possible in Matlab? I couldn't find any information or examples in the documentation.
  2 Comments
Image Analyst
Image Analyst on 22 Apr 2017
Do you mean that you want to create a PNG file with transparent pixels? If so, clarify which pixels are to be transparent - Saturn, or the space? Because when you say "the image itself", well, that is the entire image - both Saturn and the space. They're both part of the image.
Sordin
Sordin on 23 Apr 2017
I want the space to be transparent. The Saturn must be completely opaque. I am working on stacking slices to make a volumetric representation of medical data. Being able to exclude the background of each image (i.e., the parts of the image that contain no useful data) would help a lot.

Sign in to comment.

Accepted Answer

Image Analyst
Image Analyst on 23 Apr 2017
When you say "Being able to exclude the background of each image (i.e., the parts of the image that contain no useful data) would help a lot." It sounds like you want masking. So if the unwanted stuff was segmented so that it's white (like the Saturn in your binary image), then you can exclude that from the original image by masking like this:
% Mask the image using bsxfun() function
maskedRgbImage = bsxfun(@times, rgbImage, cast(mask, 'like', rgbImage));
that will give you a color image of the space/universe while the Saturn pixels would be completely black ("removed" from the image). Invert the mask using mask=~mask if you want the opposite stuff blacked out.
% Mask the image using bsxfun() function
maskedRgbImage = bsxfun(@times, rgbImage, cast(~mask, 'like', rgbImage));
If it's a simple grayscale image, you can erase mask pixels like this:
maskedGrayImage = grayImage; % Initialize
maskedGrayImage(mask) = 0;
  2 Comments
Sordin
Sordin on 24 Apr 2017
Thank you very much, the last one for grayscale images works perfectly. However, I get the following error using your code for an RGB image:
Error using cast
Too many input arguments.
Are you sure that is the correct syntax?
Walter Roberson
Walter Roberson on 24 Apr 2017
You are using an older MATLAB that does not support that syntax. You can replace
cast(~mask, 'like', rgbImage)
with
cast(~mask, class(rgbImage))

Sign in to comment.

More Answers (2)

KSSV
KSSV on 22 Apr 2017
I = imread('saturn.jpg') ;
I = rgb2gray(I) ;
I(I<50) = 255 ;
imshow(I)

Walter Roberson
Walter Roberson on 23 Apr 2017
I = imread('saturn.png');
a = im2bw(I, graythresh(I));
image(I, 'alphadata', double(a) )
Alpha data needs to be between 0 and 1 (inclusive), so just double() the logical data to get suitable alpha data.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!