How do I compute RMSE for Color Image ??

Question

Chidiebere Ike 2018-11-16

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/430256-how-do-i-compute-rmse-for-color-image

评论： Rik 2018-11-19

采纳的回答： Rik

在 MATLAB Online 中打开

I wish to compute RMSE for color images and I applied the code below was to compute RMSE for grayscale image.

Please will this same code be applicable to (RGB) color image ??

% RMSE for Grayscale Images
data = imread('woman_GT[1-Original].bmp'); % Read real data
estimate = imread('woman_GT[8-Our Method].bmp'); % Read predicted data
rmseResults=RMSE(data,estimate); % Compute RMSE
% Alert user of the answer.
message = sprintf('The Root mean square error is %.3f.', rmseResults);
msgbox(message);

Thanks

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Rik 2018-11-17

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/430256-how-do-i-compute-rmse-for-color-image#answer_347486

If the function you are using is from this FEX submission, then yes, it will work. The interpretation of the RMSE might be more difficult, or it might not even make sense.

The root mean square error is just that: sqrt(mean(err^2)), so as long as you modify that to support vectors or matrices, it should work.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Image Analyst 2018-11-19

在 MATLAB Online 中打开

Which would mean adding a dot before the caret

r = sqrt(mean(err .^ 2))

Or you could use the built-in functions immse() and psnr().

Rik 2018-11-19

I agree with the use of built-in functions whenever you have access to them, you never know where Mathworks engineers were able to put some tricks to increase performance (now or in a future release).

However, the code here will only yield the expected result for vector input, because mean will return an array with one dimension less than the input, so a 2D input will become a vector. (of course for a scalar input, the output will not be 0 dimensions)

请先登录，再进行评论。