def test_downsample_segmentation_4x_z():
case1 = np.array([ [ 0, 1 ], [ 2, 3 ] ]).reshape((2,2,1,1)) # all different
case2 = np.array([ [ 0, 0 ], [ 2, 3 ] ]).reshape((2,2,1,1)) # two are same
case3 = np.array([ [ 1, 1 ], [ 2, 2 ] ]).reshape((2,2,1,1)) # two groups are same
case4 = np.array([ [ 1, 2 ], [ 2, 2 ] ]).reshape((2,2,1,1)) # 3 are the same
case5 = np.array([ [ 5, 5 ], [ 5, 5 ] ]).reshape((2,2,1,1)) # all are the same
is_255_handled = np.array([ [ 255, 255 ], [ 1, 2 ] ], dtype=np.uint8).reshape((2,2,1,1))
test = lambda case: downsample.countless2d(case)[0][0][0][0]
assert test(case1) == 3 # d
assert test(case2) == 0 # a==b
assert test(case3) == 1 # a==b
assert test(case4) == 2 # b==c
assert test(case5) == 5 # a==b
assert test(is_255_handled) == 255
assert downsample.countless2d(case1).dtype == case1.dtype
# 0 0 1 3
# 1 1 6 3 => 1 3
case_odd = np.array([
[
[ [1] ],
[ [0] ]
],
[
[ [1] ],
[ [6] ],
],
[
[ [3] ],
[ [3] ],
],
]) # all are the same
downsamplefn = downsample.method('segmentation')
result = downsamplefn(case_odd, (2,2,1))
assert np.array_equal(result, np.array([
[
[ [1] ]
],
[
[ [3] ]
]
]))
data = np.ones(shape=(1024, 511, 62, 1), dtype=int)
result = downsamplefn(data, (2,2,1))
assert result.shape == (512, 256, 62, 1)
def downsample_and_upload(image,
bounds,
vol,
ds_shape,
mip=0,
axis='z',
skip_first=False):
ds_shape = min2(vol.volume_size, ds_shape[:3])
# sometimes we downsample a base layer of 512x512
# into underlying chunks of 64x64 which permits more scales
underlying_mip = (mip + 1) if (mip + 1) in vol.available_mips else mip
underlying_shape = vol.mip_underlying(underlying_mip).astype(np.float32)
toidx = {'x': 0, 'y': 1, 'z': 2}
preserved_idx = toidx[axis]
underlying_shape[preserved_idx] = float('inf')
# Need to use ds_shape here. Using image bounds means truncated
# edges won't generate as many mip levels
fullscales = downsample_scales.compute_plane_downsampling_scales(
size=ds_shape,
preserve_axis=axis,
max_downsampled_size=int(min(*underlying_shape)),
)
factors = downsample.scale_series_to_downsample_factors(fullscales)
if len(factors) == 0:
print(
"No factors generated. Image Shape: {}, Downsample Shape: {}, Volume Shape: {}, Bounds: {}"
.format(image.shape, ds_shape, vol.volume_size, bounds))
downsamplefn = downsample.method(vol.layer_type)
vol.mip = mip
if not skip_first:
vol[bounds.to_slices()] = image
new_bounds = bounds.clone()
for factor3 in factors:
vol.mip += 1
image = downsamplefn(image, factor3)
new_bounds //= factor3
new_bounds.maxpt = new_bounds.minpt + Vec(*image.shape[:3])
vol[new_bounds.to_slices()] = image