. Author manuscript; available in PMC: 2018 Mar 1.

Published in final edited form as: Pattern Recognit. 2016 Sep 22;63:710–718. doi: 10.1016/j.patcog.2016.09.031

Algorithm 1.

CREVER

Inputs: Grayscale image I, ρ₁, and ρ₂

Initialization:

1:
m_s ← Median of I in s × s × s kernels
2:
m_l ← Median of I in l × l × l kernels
3:
Γ₀ ← 2 – means clustered $(\frac{m_{s}}{m_{l}})$ , m_l ≠ 0
4:
Γ₀ ← Γ₀ ∪ {(x, y, z)|m_s(x, y, z) = max(m_s)}
5:
Sequence: δ₁ = max(m_s), δ₂, …, δ_i = δ_i−1 – 1, …, δ_N = min(m_s)
6:
n ← 1
7:
$I_{g} = \frac{\nabla m_{s}}{∣ ∣ \nabla m_{s} ∣ ∣}$
8:
while n < N do
9:
Γ_n ← Γ_n−1 ∪ {(x, y, z)|m_s(x, y, z) ≥ δ_n+1, ||Γ_n−1, (x, y, z)|| ≤ $\sqrt{3}$ , |Ig (x, y, z)| < τ}
10:
Γ_n ← (Γ_n ⊕ sph₁) ⊖ sph₁
11:
n ← n + 1
12:
end
13:
I_B ← Γ_N-1 > 0

Output: Binary Image I_B