Skip to content

Commit

Permalink
clean function to build stack
Browse files Browse the repository at this point in the history
  • Loading branch information
@Henley13
Henley13 committed May 5, 2020
1 parent 375df0b commit 8fc4b38
Showing 1 changed file with 34 additions and 80 deletions.
114 changes: 34 additions & 80 deletions bigfish/stack/preprocess.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -24,9 +24,8 @@

# ### Building stack ###

def build_stacks(data_map, input_dimension=None, check=False, normalize=False,
channel_to_stretch=None, stretching_percentile=99.9,
cast_8bit=False, return_origin=False):
def build_stacks(data_map, input_dimension=None, sanity_check=False,
return_origin=False):
"""Generator to build several stacks from recipe-folder pairs.
To build a stack, a recipe should be linked to a directory including all
Expand Down Expand Up @@ -104,24 +103,15 @@ def build_stacks(data_map, input_dimension=None, check=False, normalize=False,
Map between input directories and recipes.
input_dimension : int
Number of dimensions of the loaded files.
check : bool
sanity_check : bool
Check the validity of the loaded tensor.
normalize : bool
Normalize the different channels of the loaded stack (rescaling).
channel_to_stretch : int or List[int]
Channel to stretch.
stretching_percentile : float
Percentile to determine the maximum intensity value used to rescale
the image.
return_origin : bool
Return the input directory and the recipe used to build the stack.
cast_8bit : bool
Cast tensor in np.uint8.
Returns
-------
tensor : np.ndarray, np.uint
Tensor with shape (r, c, z, y, x).
tensor : np.ndarray
Tensor with shape (round, channel, z, y, x).
input_directory : str
Path of the input directory from where the tensor is built.
recipe : dict
Expand All @@ -130,6 +120,8 @@ def build_stacks(data_map, input_dimension=None, check=False, normalize=False,
"""
# check parameters
check_parameter(data_map=list,
input_dimension=(int, type(None)),
sanity_check=bool,
return_origin=bool)

# load and generate tensors for each recipe-folder pair
Expand All @@ -139,17 +131,15 @@ def build_stacks(data_map, input_dimension=None, check=False, normalize=False,
nb_fov = count_nb_fov(recipe)
for i_fov in range(nb_fov):
tensor = build_stack(recipe, input_folder, input_dimension, i_fov,
check, normalize, channel_to_stretch,
stretching_percentile, cast_8bit)
sanity_check)
if return_origin:
yield tensor, input_folder, recipe, i_fov
else:
yield tensor


def build_stack(recipe, input_folder, input_dimension=None, i_fov=0,
check=False, normalize=False, channel_to_stretch=None,
stretching_percentile=99.9, cast_8bit=False):
sanity_check=False):
"""Build 5-d stack and normalize it.
The recipe dictionary for one field of view takes the form:
Expand Down Expand Up @@ -217,51 +207,34 @@ def build_stack(recipe, input_folder, input_dimension=None, i_fov=0,
Number of dimensions of the loaded files.
i_fov : int
Index of the fov to build.
check : bool
sanity_check : bool
Check the validity of the loaded tensor.
normalize : bool
Normalize the different channels of the loaded stack (rescaling).
channel_to_stretch : int or List[int]
Channel to stretch.
stretching_percentile : float
Percentile to determine the maximum intensity value used to rescale
the image.
cast_8bit : bool
Cast the tensor in np.uint8.
Returns
-------
tensor : np.ndarray, np.uint
Tensor with shape (r, c, z, y, x).
tensor : np.ndarray
Tensor with shape (round, channel, z, y, x).
"""
# check parameters
check_recipe(recipe)
check_parameter(input_folder=str,
input_dimension=(int, type(None)),
i_fov=int,
check=bool,
normalize=bool,
channel_to_stretch=(int, list, type(None)),
stretching_percentile=float,
cast_8bit=bool)
sanity_check=bool)

# build stack from recipe and tif files
tensor = _load_stack(recipe, input_folder, input_dimension, i_fov)

# check the validity of the loaded tensor
if check:
if sanity_check:
check_array(tensor,
dtype=[np.uint8, np.uint16, np.uint32,
np.int8, np.int16, np.int32,
np.float16, np.float32, np.float64,
bool],
ndim=5,
dtype=[np.uint8, np.uint16])

# rescale data and improve contrast
if normalize:
tensor = rescale(tensor, channel_to_stretch, stretching_percentile)

# cast in np.uint8 if necessary, in order to reduce memory allocation
if tensor.dtype == np.uint16 and cast_8bit:
tensor = cast_img_uint8(tensor)
allow_nan=False)

return tensor

Expand Down Expand Up @@ -573,61 +546,42 @@ def _get_input_dimension(recipe, input_folder):
return nb_dim


def build_stack_no_recipe(paths, input_dimension=None, check=False,
normalize=False, channel_to_stretch=None,
stretching_percentile=99.9, cast_8bit=False):
"""Build 5-d stack and normalize it, without recipe.
def build_stack_no_recipe(paths, input_dimension=None, sanity_check=False):
"""Build 5-d stack without recipe.
Parameters
----------
paths : List[str]
List of the paths to stack.
input_dimension : str
Number of dimensions of the loaded files.
check : bool
Check the validity of the loaded tensor.
normalize : bool
Normalize the different channels of the loaded stack (rescaling).
channel_to_stretch : int or List[int]
Channel to stretch.
stretching_percentile : float
Percentile to determine the maximum intensity value used to rescale
the image.
cast_8bit : bool
Cast the tensor in np.uint8.
sanity_check : bool
Check the validity of the loaded tensor
Returns
-------
tensor : np.ndarray, np.uint
Tensor with shape (r, c, z, y, x).
tensor : np.ndarray
Tensor with shape (round, channel, z, y, x).
"""
# check parameters
check_parameter(paths=(str, list),
input_dimension=(int, type(None)),
normalize=bool,
channel_to_stretch=(int, list, type(None)),
stretching_percentile=float,
cast_8bit=bool)
sanity_check=bool)

# build stack from tif files
tensor = _load_stack_no_recipe(paths, input_dimension)

# check the validity of the loaded tensor
if check:
if sanity_check:
check_array(tensor,
dtype=[np.uint8, np.uint16, np.uint32,
np.int8, np.int16, np.int32,
np.float16, np.float32, np.float64,
bool],
ndim=5,
dtype=[np.uint8, np.uint16],
allow_nan=False)

# rescale data and improve contrast
if normalize:
tensor = rescale(tensor, channel_to_stretch, stretching_percentile)

# cast in np.uint8 if necessary, in order to reduce memory allocation
if tensor.dtype == np.uint16 and cast_8bit:
tensor = cast_img_uint8(tensor)

return tensor


Expand All @@ -647,7 +601,7 @@ def _load_stack_no_recipe(paths, input_dimension=None):
Returns
-------
tensor_5d : np.ndarray, np.uint
Tensor with shape (r, c, z, y, x).
Tensor with shape (round, channel, z, y, x).
"""
# load an image and get the number of dimensions
Expand All @@ -674,8 +628,8 @@ def _load_stack_no_recipe(paths, input_dimension=None):
tensor_5d = stacks[0]
else:
raise ValueError("Files do not have the right number of dimensions: "
"{0}. The files we stack should be in 2-d, 3-d, 4-d "
"or 5-d.".format(input_dimension))
"{0}. The files we stack should have between 2 and "
"5 dimensions.".format(input_dimension))

return tensor_5d

Expand Down

0 comments on commit 8fc4b38

Please sign in to comment.