2D Phase unwrapping using Minimum Cost Flow (MCF)

2D unwrappers can be easily built by combining the MCF solvers and planar graph interfaces available in spurt. In the examples below, we use the OR Tools solver but the interface classes can be used to add support for other MCF solvers as well.

Regular grid 2D MCF

A regular 2D grid unwrapper can be implemented as follows using spurt

# igram is a 2D array of np.complex64
g = spurt.graph.Reg2DGraph(igram.shape)
solver = spurt.mcf.ORMCFSolver(g)

# Assume unit cost
cost = np.ones(solver.edges.shape[0], dtype=int)
unw, _ = solver.unwrap_one(igram.flatten(), cost)
unw = unw.reshape(igram.shape)

Irregular grid 2D MCF

A sparse 2D grid unwrapper can be implemented as follows using spurt

# igram is a 1D array of type np.complex64 and length npts
# xy is a npts x 2 array with coordinates of pixels

graph = spurt.graph.DelaunayGraph(xy)
solver = spurt.mcf.ORMCFSolver(graph)

# Use unit cost
cost = np.ones(solver.edges.shape[0], dtype=int)
unw, _ = solver.unwrap_one(igram, cost)