
170 Variational Methods in Image Processing
FIGURE 8.20: Joint segmentation, denoising, and deblurring using the
binary level set model. Top row: degraded image g (blurred with motion
blur kernel of length 10, oriented at an angle θ = 25
◦
w.r.t. the horizon
and contaminated by Gaussian noise with σ
n
= 10), orig inal image. Rows
2 through 5: initial curves, curve evolution at iterations 50, 100, 300 with
µ = 5 · 2 55
2
, and the restore d imag e u (SNR = 28.1827). (c
1
, c
2
): original
image ≈ (62.7525, 25 9.8939 ), restored u, (61.9194, 262.7795).
8.4 Exercises
Exercise 8.1 Prove the validity of formulas (8.2) and (8.3).
Exercise 8.2 Assume that for t ≥ 0,