
456 A Computational Introduction to Digital Image Processing, Second Edition
15.7 Denoising Using Wavelets
Since noise can be considered a high frequency component, some form of low pass fil-
tering should remove it. As we did for compression, we can remove such components by
thresholding. To compare with the results obtained in Chapter 8, we shall use a gray version
of the seagull image, converted to type
double
.
MATLAB/Octave
>> f = im2double(rgb2gray(imread(’gull.png’)));
>> fg = imnoise(f,’gaussian’);
>> imshow(fg)
Now we can attempt some noise removal. We shall assume that all the filter co efficients
are those of the Daubechies-4 wavelet.
MATLAB/Octav ...