def __init__(self, *args, **kwargs):
SceneProcess.__init__(self, *args, **kwargs)
sc = self.scene
self.original = ADict()
self.original.update({
"displacement": sc.displacement.copy(),
"theta": sc.theta.copy(),
"phi": sc.phi.copy(),
"frame.dLat": sc.frame.dLat,
"frame.dLon": sc.frame.dLon,
})
def __init__(self, *args, **kwargs):
SceneProcess.__init__(self, *args, **kwargs)
sc = self.scene
self.original = ADict()
self.original.update({
'displacement': sc.displacement.copy(),
'theta': sc.theta.copy(),
'phi': sc.phi.copy(),
'frame.dLat': sc.frame.dLat,
'frame.dLon': sc.frame.dLon,
})
class Downsample(SceneProcess):
from scipy import ndimage
factor = Int.T(help="Downsample factor.")
def __init__(self, *args, **kwargs):
SceneProcess.__init__(self, *args, **kwargs)
sc = self.scene
self.original = ADict()
self.original.update(
{
"displacement": sc.displacement.copy(),
"theta": sc.theta.copy(),
"phi": sc.phi.copy(),
"frame.dLat": sc.frame.dLat,
"frame.dLon": sc.frame.dLon,
}
)
def apply(self):
sc = self.scene
org = self.original
factor = self.factor
sx, sy = sc.displacement.shape
gx, gy = num.ogrid[0:sx, 0:sy]
regions = sy / factor * (gx / factor) + gy / factor
indices = num.arange(regions.max() + 1)
def block_downsample(arr):
res = ndimage.mean(arr, labels=regions, index=indices)
res.shape = (sx / factor, sy / factor)
return res
sc.displacement = block_downsample(sc.displacement)
sc.theta = block_downsample(sc.theta)
sc.phi = block_downsample(sc.phi)
sc.frame.dLat = org["frame.dLat"] * self.factor
sc.frame.dLon = org["frame.dLat"] * self.factor
def remove(self):
for prop, value in self.original.items():
self.scene.__setattr__(prop, value)
del self.original