Create composite mask, and apply this mask to an image

Hi,

Scanning samples using NIR, processing the spectra in matlab.

I have an image, and I need to create a mask to remove the background.

I have been instructed to snip sections of the image, creating masks.

I then have to combine these masks (x3) into one composite and apply it to the overall image. Removing the background.

CL below is a crop of image I5.

CL=I5(131:149,36:43,:);

[xC,yC,zC]=size(CL);

figure,imshow(mean(CL,3),[])

I searched through a loop showing each slice of the hypercube. Slice for example 69 has contrast between the item and the background.

The threshold excludes the background.

%%CL snip one

figure,

subplot(1,2,1),imshow(CL(:,:,69)<0.26,[])

subplot(1,2,2),histogram(unfold(CL(:,:,69)),100)

CL_mask=CL(:,:,69)<0.26;

%% CL3 my second crop of the image

CL3=I5(125:150, 15:34,:);

[xC3,yC3,zC3]=size(CL3);

figure,imshow(mean(CL3,3),[])

CL_mask3=CL3(:,:,81)<0.1;

%% CL4 snip four

CL4=I5(153:168,25:41,:);

[xC4,yC4,zC4]=size(CL4);

figure,imshow(mean(CL4,3),[])

CL_mask4=CL4(:,:,65)<0.24;

Is there a way to combine the three masks? CL_mask4 + CLmask3 + CL_mask ?

I am a beginner level user, but I can provide more info if it helps.

Help would be hugely appreciated.

Shane

Answer 1

[Tarunbir Gambhir]

A better approach would be to use a single mask for your entire image. Start with a matrix of same size as your image and update it at locations where the threshold condition suffices inside the cropped regions.

Something like this could be done:

% Assuming the dimensions of image to be Spacial x Spacial x Spectral
mask = zeros(size(I5,[1,2])); % Initialize mask with zero across spacial dimensions

%% CL 
mask(131:149,36:43)=I5(131:149,36:43,69)<0.26;

%% CL3 
mask(125:150,15:34)=I5(125:150,15:34,81)<0.1;

%% CL4 
mask(153:168,25:41)=I5(153:168,25:41,65)<0.24;

Repeating the mask value across all spectral slices and then dot-multiplication with the original image should give you the masked image.

%% Applying the mask to I5
final_image = I5 .* repmat(mask,[1 1 size(I5,3)]); 

[Shane Browne]

Thank you very much for the help it is greatly appreciated.

The image size was defined as:

[x,y,z]=size(I5);

I previously had a loop to check if a mask worked on all slices of the hypercube.

CL_mask=CL(:,:,69)<0.26;

CL_masked=zeros(size(CL));

for i=1:z

CL_masked(:,:,i)=CL(:,:,i).*CL_mask;

end

%taking a look through the slices to check the mask worked ok

figure,

for i=1:z

imshow(CL_masked(:,:,i),[]),title(sprintf('%d',i))

pause (0.1)

end

Is it possible to run a loop like this checking each slice with the new mask?

[Tarunbir Gambhir]

Yes. After getting the mask you could use a "for" loop in this way to check if the mask worked for every slice.