error in std for 3d matrix

Question

Anastasia 2016-12-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/316582-error-in-std-for-3d-matrix

评论： Guillaume 2016-12-13

Hey guys!

I am trying to find the std of a 3D matrix, where size(X) is 207 x 201 x 34. Since I must reshape to find std, I am using this formula: std(reshape(X,1,207*201*34)).

And I obtain this error: Error in std (line 31) y = sqrt(var(varargin{:}));

When I look at the syntax, I thought of using std(reshape(X,1,207*201*34), 0, 3), where 3 is for 3D, however I still obtain the same error.

What I am missing?

Thank you very much in advance!

2 个评论
显示无隐藏无

Steven Lord 2016-12-12

What is the full text of the error message (including everything printed in red?)

Anastasia 2016-12-13

编辑：Anastasia 2016-12-13

Hello, this is the full text of the error message:

Error using var (line 58) First argument must be single or double.

Error in std (line 31) y = sqrt(var(varargin{:}));

Error in Code3D (line 27) th=mean(mean(mean(X))) + std(reshape(X,1,207*201*34));

Is the problem that X is nor single nor double, but 207*201*34? How can I fix that?

Thank you!

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Isabella Osetinsky-Tzidaki 2016-12-12

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/316582-error-in-std-for-3d-matrix#answer_246914

try std(x(:))

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Anastasia 2016-12-13

Hello! Thank you for your answer but I obtain the same error...

请先登录，再进行评论。

Answer 2

Guillaume 2016-12-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/316582-error-in-std-for-3d-matrix#answer_246997

在 MATLAB Online 中打开

Error: First argument must be single or double

Well, clearly, your first argument (a transform of X) is not single or double. So what is:

class(X)

If it's an integer type:

Xdouble = double(X);
th = mean(Xdouble(:)) + std(Xdouble(:));

2 个评论
显示无隐藏无

Anastasia 2016-12-13

The class is uint16. I tried it and it works, is it correct to use the code you gave me with a uint16 class?

Thanks!

Guillaume 2016-12-13

Without context it's impossible to say if it's correct or not. Certainly, if you want the standard deviation of your integers, you have to convert them to double. If an integer result is then required, you can convert the output back to uint16 (with the loss of precision that it entails).

请先登录，再进行评论。