1. Space-invariant blurring Perform Gaussian blurring on Mandrill.pgm with standard deviation σ. Assume space-invariant blur and a kernel of size d6σ+1e×d6σ+1e. Observe the outputs for these values of σ: 1.6, 1.2, 1.0, 0.6, 0.3 and 0.0.
2. Space-variant blurring Now assume the blur to be space-variant, i.e. the standard deviation varies for each pixel. Consider the distribution of σ to be
´
σ(m,n)=Aexp, 0≤m,n≤N−1
B
with
µN N¶
σ , =2.0 and σ(0,0)=0.01,
2 2
where N×N is size of the image and pixel indices are in the range [0,N−1]×[0,N−1]. Find A and B, and create the matrix σ. Perform Gaussian blurring on Globe.pgm using the values of σ(m,n).
3. Blur Nautilus.pgm using
(a) space-invariant blur code of part 1 with σ=1.0, and
(b) space-variant blur code of part 2 with σ(m,n)=1.0 for 0≤m,n≤N−1.
Verify that the blurred images of the above two steps are same.
