Reply by Giovanni Marino January 30, 20102010-01-30
Dear All,
I am trying to denoising some simulated images by using a Gaussian filter. I
would check if my results are right by deconvolving the output images in the
frequency domain. Suppose that Image_out is the output of my filter, in the
frequency domain we should get: F(Image_out)=F(Image_inp)*F(Filter_Gaus),
where F() represent the Fourier transform. As a consequence the
deconvolution result should be the inverse Fourier transfer
of F(Image_out)*(F(Filter_Gaus))^{-1}, i.e. the multiplication of the
Fourier transform of the Image_out by the 1/(F(Filter_Gaus)). Now the
problem: Since the window size of Gaussian filter is smaller than image size
(i.e. image is 100x100 pixels, whereas the Gaussian filter window size is
10x10 pixels), how can implement my deconvolution? Any suggestion?
Many thanks in advance for your time.
Regards
Giovanni