Sei sulla pagina 1di 17

Image processing

Adaptive filters:
Wiener and Lucy Richardson filters
Image model

 Original data = Image * PSF of blurring +additive noise


 * - convolution operator

The Importance of the PSF


 Based on this model, the fundamental task of
deblurring is to deconvolve the blurred image with
the PSF that exactly describes the distortion.
 The quality of the deblurred image is mainly
determined by knowledge of the PSF.
Create a PSF of the blur and blur

I = imread('flowers.tif');
figure; imshow(I); title('Original Image');
LEN = 31; % length of blur
THETA = 11; % angle of blur
PSF = fspecial('motion',LEN,THETA); %create PSF
Blurred = imfilter(I,PSF,'circular','conv'); %blur
figure; imshow(Blurred);title('Blurred Image');

imshow(imabsdiff(I,Blurred); % difference
Wiener filter

 It removes the additive noise and inverts


the blurring simultaneously.
 The Wiener filtering is optimal in terms of
the mean square error.
Filtering in the Frequency Domain

1. Transform image data to the frequency


domain via the FFT
2. Multiply the image's spectrum with some
filtering mask
3. Transform the spectrum back to the spatial
domain that corresponds to deconvolution
in the time domain
Wiener filter

Where Sxx (f1,f2), Sηη(f1,f2) are respectively power


spectra of the original image and the additive noise,
and H(f1,f2) is the blurring filter.
Wiener filter has two separate parts, an inverse
filtering part and a noise smoothing part.
 It performs the deconvolution by inverse filtering
(highpass filtering) and removes the noise with a
compression operation (lowpass filtering).
Power Spectrum

 The most common way of generating a


power spectrum is by using a Fourier
transform and taking the square of the
magnitude of the complex coefficients.
 Other techniques such as the maximum
entropy method can also be used to
estimate the power spectrum.
Implementation
 To implement the Wiener filter in practice we have to
estimate the power spectra of the original image and
the additive noise.
 For white additive noise the power spectrum is equal
to the variance of the noise.
 To estimate the power spectrum of the original image
many methods can be used. A direct estimate is the
estimate of the power spectrum computed from the
observation:

 Where Y(k,l) is the DFT of the observation.


Discrete Fourier Transform
deblur the image by deconvolving

wnr1 = deconvwnr(Blurred,PSF);
figure;imshow(wnr1);
title('Restored, True PSF');
Another blurring
Lucy-Richardson
 Lucy-Richardson
denoising accounts on
photon counting noise
with a Poisson
distribution

 There is also as you


know the readout
noise of Gaussian
distribution
Lucy-Richardson
 The Lucy-Richardson algorithm generates a
restored image through an iterative
method.
 The essence of the iteration is as follows:
the (n+1)th estimate of the restored image
is given by the nth estimate of the restored
image multiplied by a correction image.
That is,
original data
image = image --------------- ● PSF*
n+1 n image * PSF
n
Implementation
luc1 = deconvlucy(Blurred,PSF,5);
figure; imshow(luc1);
title('Restored Image, NUMIT = 5');

 deconvlucy function, by default, performs multiple iterations of


the deblurring process. You can stop the processing, after a
certain number of iterations, to check the result, and then
restart the iterations from the point where processing stopped.

The output cell array contains these four elements:


output{1} -- The original input image
output{2} -- The image produced by the last iteration
output{3} -- The image produced by the next-to-last iteration
output{4} -- Internal information used by deconvlucy to know
where to restart the process
Result of 5 iterations
Result of 10 iterations
Image difference
 imabsdiff

Potrebbero piacerti anche