Is there a built-in Demosaicing option in Matlab which doesn't use interpolation and instead just reduces the image size?

Question

Shida 2017-8-17

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/353074-is-there-a-built-in-demosaicing-option-in-matlab-which-doesn-t-use-interpolation-and-instead-just-re

评论： MatlabUser 2018-10-19

I am aware of demosaic(I,sensorAlignment) function in matlab. But if one would want to reduce noise and just reduce the resolution instead (e.g, to a quarter), does matlab already have a built-in function for that?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

MatlabUser 2018-10-19

what is the relation between size and interpolation ?

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Neil Guertin 2017-8-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/353074-is-there-a-built-in-demosaicing-option-in-matlab-which-doesn-t-use-interpolation-and-instead-just-re#answer_278863

在 MATLAB Online 中打开

There is not currently a built in way to do this. However, it is pretty simple to write:

 I = imread('mandi.tif');
 I = I(1:end-mod(end,2),1:end-mod(end,2)); %In case image dimensions are odd
 I2 = zeros([size(I)/2 3],class(I));
 I2(:,:,1) = I(2:2:end,2:2:end);
 I2(:,:,2) = I(1:2:end,2:2:end)/2 + I(2:2:end,1:2:end)/2;
 I2(:,:,3) = I(1:2:end,1:2:end);
 imshow(I2)

However, there will be some artifacts that interpolation would have smoothed out: look at Mandi's chin for example. Also note that you are actually losing resolution in the green channel by averaging the two green pixels when you do it this way.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Shida 2017-8-23

Thanks Neil. You are right, interpolation does help against artifacts. I am using a Canon EOS with .cr2 format so I have an 'rggb' bayer pattern. But indeed I don't get much difference if I skip interpolation (and instead just reduce the image size to a quarter). I am using dcraw for conversion which seems to already do a reasonable job and I wasn't able to get anything better this way. Thanks again for your response.

请先登录，再进行评论。