288 ◾ David R. Burton
where Δ
x
ϕ(i,j) and Δ
y
ϕ(i, j) are the unwrapped phase gradients in the x and
y directions, respectively, given by
∆φ =φ +−φ(, )(1, )(
ij ij
x
(8.6)
and
∆φ =φ +−φ(, )(,1)(
ij ij
y
(8.7)
e wrapped phase derivatives are similarly dened on the wrapped data.
From this, it can be seen that, although seemingly a dierent approach
than those we have seen previously, in reality this, at its most fundamental
level, is a dierencing operation like the original algorithm we dened.
e exact form of the minimizing equation depends on which minimi-
zation technique is employed. Most success has been achieved using one
of three major approaches:
• Unweighted least squares: First introduced by Hunt in 1979 [14], in
this technique ...