Skip to content

Commit

Permalink
add tests for filtering
Browse files Browse the repository at this point in the history
  • Loading branch information
@Henley13
Henley13 committed May 5, 2020
1 parent 98d8ffd commit f95ed8c
Show file tree
Hide file tree
Showing 2 changed files with 474 additions and 67 deletions.
127 changes: 60 additions & 67 deletions bigfish/stack/filter.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,14 +4,24 @@

import numpy as np

from .utils import check_array, check_parameter
from .preprocess import (cast_img_float32, cast_img_float64, cast_img_uint8,
cast_img_uint16)

from skimage.morphology.selem import square, diamond, rectangle, disk
from skimage.morphology import (binary_dilation, dilation, binary_erosion,
erosion)
from skimage.filters import rank, gaussian
from .utils import check_array
from .utils import check_parameter

from .preprocess import cast_img_float32
from .preprocess import cast_img_float64
from .preprocess import cast_img_uint8
from .preprocess import cast_img_uint16

from skimage.morphology.selem import square
from skimage.morphology.selem import diamond
from skimage.morphology.selem import rectangle
from skimage.morphology.selem import disk
from skimage.morphology import binary_dilation
from skimage.morphology import dilation
from skimage.morphology import binary_erosion
from skimage.morphology import erosion
from skimage.filters import rank
from skimage.filters import gaussian

from scipy.ndimage import gaussian_laplace

Expand All @@ -28,7 +38,7 @@ def _define_kernel(shape, size, dtype):
'rectangle' or 'square').
size : int, Tuple(int) or List(int)
The size of the kernel:
- For the rectangle we expect two values (width, height).
- For the rectangle we expect two values (height, width).
- For the square one value (width).
- For the disk and the diamond one value (radius).
dtype : type
Expand All @@ -50,7 +60,9 @@ def _define_kernel(shape, size, dtype):
elif shape == "square":
kernel = square(size, dtype=dtype)
else:
raise ValueError("Kernel definition is wrong.")
raise ValueError("Kernel definition is wrong. Shape of the kernel "
"should be 'diamond', 'disk', 'rectangle' or "
"'square'. Not {0}.".format(shape))

return kernel

Expand All @@ -67,7 +79,7 @@ def mean_filter(image, kernel_shape, kernel_size):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
Expand Down Expand Up @@ -105,7 +117,7 @@ def median_filter(image, kernel_shape, kernel_size):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
Expand Down Expand Up @@ -143,7 +155,7 @@ def maximum_filter(image, kernel_shape, kernel_size):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
Expand Down Expand Up @@ -181,7 +193,7 @@ def minimum_filter(image, kernel_shape, kernel_size):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
Expand All @@ -207,7 +219,7 @@ def minimum_filter(image, kernel_shape, kernel_size):
return image_filtered


def log_filter(image, sigma, keep_dtype=False):
def log_filter(image, sigma):
"""Apply a Laplacian of Gaussian filter to a 2-d or 3-d image.
The function returns the inverse of the filtered image such that the pixels
Expand All @@ -222,8 +234,6 @@ def log_filter(image, sigma, keep_dtype=False):
sigma : float, int, Tuple(float, int) or List(float, int)
Sigma used for the gaussian filter (one for each dimension). If it's a
float, the same sigma is applied to every dimensions.
keep_dtype : bool
Cast output image as input image.
Returns
-------
Expand Down Expand Up @@ -259,36 +269,33 @@ def log_filter(image, sigma, keep_dtype=False):
image_filtered = np.clip(-image_filtered, a_min=0, a_max=None)

# cast filtered image
if keep_dtype:
if image.dtype == np.uint8:
image_filtered = cast_img_uint8(image_filtered)
elif image.dtype == np.uint16:
image_filtered = cast_img_uint16(image_filtered)
else:
pass
if image.dtype == np.uint8:
image_filtered = cast_img_uint8(image_filtered, catch_warning=True)
elif image.dtype == np.uint16:
image_filtered = cast_img_uint16(image_filtered, catch_warning=True)
else:
pass

return image_filtered


def gaussian_filter(image, sigma, allow_negative=False, keep_dtype=False):
def gaussian_filter(image, sigma, allow_negative=False):
"""Apply a Gaussian filter to a 2-d or 3-d image.
Parameters
----------
image : np.ndarray, np.uint
image : np.ndarray
Image with shape (z, y, x) or (y, x).
sigma : float, int, Tuple(float, int) or List(float, int)
Sigma used for the gaussian filter (one for each dimension). If it's a
float, the same sigma is applied to every dimensions.
allow_negative : bool
Allow negative values after the filtering or clip them to 0.
keep_dtype : bool
Cast output image as input image. Integer output can't allow negative
values.
Allow negative values after the filtering or clip them to 0. Not
compatible with unsigned integer images.
Returns
-------
image_filtered : np.ndarray, np.float
image_filtered : np.ndarray
Filtered image.
"""
Expand All @@ -299,12 +306,17 @@ def gaussian_filter(image, sigma, allow_negative=False, keep_dtype=False):
check_parameter(sigma=(float, int, tuple, list),
allow_negative=bool)

if image.dtype in [np.uint8, np.uint16] and allow_negative:
raise ValueError("Negative values are impossible with unsigned "
"integer image.")

# we cast the data in np.float to allow negative values
image_float = None
if image.dtype == np.uint8:
image_float = cast_img_float32(image)
elif image.dtype == np.uint16:
image_float = cast_img_float64(image)
else:
image_float = image

# we apply gaussian filter
image_filtered = gaussian(image_float, sigma=sigma)
Expand All @@ -314,13 +326,12 @@ def gaussian_filter(image, sigma, allow_negative=False, keep_dtype=False):
image_filtered = np.clip(image_filtered, a_min=0, a_max=None)

# cast filtered image
if keep_dtype and not allow_negative:
if image.dtype == np.uint8:
image_filtered = cast_img_uint8(image_filtered)
elif image.dtype == np.uint16:
image_filtered = cast_img_uint16(image_filtered)
else:
pass
if image.dtype == np.uint8:
image_filtered = cast_img_uint8(image_filtered, catch_warning=True)
elif image.dtype == np.uint16:
image_filtered = cast_img_uint16(image_filtered, catch_warning=True)
else:
pass

return image_filtered

Expand All @@ -330,30 +341,21 @@ def remove_background_mean(image, kernel_shape="disk", kernel_size=200):
Parameters
----------
image : np.ndarray, np.uint8
Image to process with shape (y, x). Casting in np.uint8 makes the
computation faster.
image : np.ndarray, np.uint
Image to process with shape (y, x).
kernel_shape : str
Shape of the kernel used to compute the filter ('diamond', 'disk',
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
image_without_back : np.ndarray, np.uint
Image processed.
"""
# check parameters
check_array(image,
ndim=2,
dtype=[np.uint8])
# TODO allow np.uint16 ?
check_parameter(kernel_shape=str,
kernel_size=(int, tuple, list))

# compute background noise with a large mean filter
background = mean_filter(image,
kernel_shape=kernel_shape,
Expand Down Expand Up @@ -387,16 +389,9 @@ def remove_background_gaussian(image, sigma):
Image processed with shape (z, y, x) or (y, x).
"""
# check parameters
check_array(image,
ndim=[2, 3],
dtype=[np.uint8, np.uint16, np.float32, np.float64])
check_parameter(sigma=(float, int, tuple, list))

# apply a gaussian filter
image_filtered = gaussian_filter(image, sigma,
allow_negative=False,
keep_dtype=True)
allow_negative=False)

# substract the gaussian filter
out = np.zeros_like(image)
Expand All @@ -420,19 +415,18 @@ def dilation_filter(image, kernel_shape=None, kernel_size=None):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
image_filtered : np.ndarray, np.uint
image_filtered : np.ndarray
Filtered 2-d image with shape (y, x).
"""
# TODO check dtype
# check parameters
check_array(image,
ndim=2,
dtype=[np.uint8, np.uint16, bool])
dtype=[np.uint8, np.uint16, np.float32, np.float64, bool])
check_parameter(kernel_shape=(str, type(None)),
kernel_size=(int, tuple, list, type(None)))

Expand Down Expand Up @@ -465,19 +459,18 @@ def erosion_filter(image, kernel_shape=None, kernel_size=None):
'rectangle' or 'square').
kernel_size : int or Tuple(int)
The size of the kernel. For the rectangle we expect two integers
(width, height).
(height, width).
Returns
-------
image_filtered : np.ndarray, np.uint
image_filtered : np.ndarray
Filtered 2-d image with shape (y, x).
"""
# TODO check dtype
# check parameters
check_array(image,
ndim=2,
dtype=[np.uint8, np.uint16, bool])
dtype=[np.uint8, np.uint16, np.float32, np.float64, bool])
check_parameter(kernel_shape=(str, type(None)),
kernel_size=(int, tuple, list, type(None)))

Expand Down
Loading

0 comments on commit f95ed8c

Please sign in to comment.