Error using imgaussfilt Expected A to be real.

Question

Gulfam Saju 2022-3-31

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1685694-error-using-imgaussfilt-expected-a-to-be-real

回答： Paras Gupta 2023-9-22

%demo
clear all;
CoilNum = 12;
load slice66.mat;
for s = 1 : CoilNum
    Img(:,:,s) = ifft2(raw_data(:,:,s));
end
image1= Img;
load slice94.mat;
for s = 1 : CoilNum
    Img2(:,:,s) = ifft2(raw_data(:,:,s));
end
image2= Img2;
psnr_=getPSNR(image1,image2);
ssim_=getMSSIM(image1,image2);
fprintf('PSNR= %f - SSIM= %f\n',psnr_,ssim_);

While I run this code, I get this error.

Error using imgaussfilt
Expected A to be real.
Error in imgaussfilt>parseInputs (line 398)
validateattributes(A, supportedClasses, supportedImageAttributes, mfilename, 'A');
Error in imgaussfilt (line 121)
[A, options] = parseInputs(args{:});
Error in getMSSIM (line 16)
mu1=imgaussfilt(frameReference,1.5);
Error in demo (line 16)
ssim_=getMSSIM(image1,image2);

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

Paras Gupta 2023-9-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1685694-error-using-imgaussfilt-expected-a-to-be-real#answer_1316032

在 MATLAB Online 中打开

Hi Gulfam,

I understand that you are getting an error while using the “imgaussfilt” function in MATLAB.

The error message in the question indicates that the “imgaussfilt” function expects the input image A to be real, but it seems to be receiving a complex-valued image.

The potential reason for this issue is the use of “ifft2” function on the variable “raw_data” for each of the slices. Fourier transform operates in the frequency domain, where the data can have both real and imaginary components. When performing an inverse Fourier transform using “ifft2” function in MATLAB, the resulting image may be complex-valued. This can happen in the following cases:

If the input is complex-valued
If the input data is not symmetric or centered
Even when the input data is real-valued, the output may have small imaginary components that are close to zero due to numerical precision limitations.

To resolve this issue, you can use the ‘real’ function in MATLAB to extract the real part of the complex valued images before passing them to the “getMSSIM” function.

image1 = real(Img); % Convert to real-valued image 
image2 = real(Img2); % Convert to real-valued image 
psnr_=getPSNR(image1,image2); 
ssim_=getMSSIM(image1,image2); 
fprintf('PSNR= %f - SSIM= %f\n',psnr_,ssim_); 