
1.In averaging de-nosing method, how much improvement has been made for SNR of averaged image?
Answer:Noise is a random variable distributed independently with the original signal,we determine form the above equation:
Then the SNR of averaged image increase n times as there are n terms of date for the average.
2.Derive the formula of averaging using discrete convolution
Answer : The discrete convolution is :
A is the matrix of image and B is the filter template,when processing a image,often we should take the edge part into consideration,sometomes 0 is added and sometimes we use mirro symmetry.
3.Program a median filter and filter a noise image
%20%salt &pepper noise is add to the original image,the filter window is the default set 3*3 .
s=imread(‘’)
f=imnoise(f,’salt & pepper’,0.2);
f1=medfilt2(f);
figure(1);imshow(s);
title(‘original image of sanmao’);
figure(2);imshow(f);
title(‘noise corrupted image’);
figure(3);imshow(f1);
title(‘noise filtered image’)
the result is blew:
%gaussion noise is another common noise,so we add it to the image:
s=imread(‘’);
f=imnoise(s,’gaussian’,0,0.01);
f1=medfilt(f);
The result is blew:
Conclusion: conventional median filter is useful to deal with salt and pepper noise,the noise density is usually blew 50 persent. However ,median filter work not as good as gaussion
