Add srm and srg tests and remove extra commenting in repo

dpm_tools/io/__init__.py

@@ -24,8 +24,8 @@
 
 """
 
-from ._io_utils import find_files_with_ext, get_tiff_metadata, natural_sort, combine_slices, convert_filetype
+from .io_utils import find_files_with_ext, get_tiff_metadata, natural_sort, combine_slices, convert_filetype
 
-from ._read_data import read_image, Image
+from .read_data import read_image, Image
 
-from ._write_data import write_image
+from .write_data import write_image

dpm_tools/io/_io_utils.py → dpm_tools/io/io_utils.py

@@ -6,8 +6,8 @@
 from typing import Any, Tuple
 import re
 
-from ._read_data import read_image
-from ._write_data import write_image
+from .read_data import read_image
+from .write_data import write_image
 
 
 def find_files_with_ext(directory: pathlib.Path, extension: str) -> list:

dpm_tools/io/_read_data.py → dpm_tools/io/read_data.py

@@ -293,66 +293,3 @@ def _read_data_from_files(self):
             self.vector = images[1:]
 
 
-# @dataclass(kw_only=True)
-# class ImageFromFile(Image):
-#     """
-#     ImageFromFile dataclass inherited from the Image dataclass. This dataclass allows users to pass the path to an image file
-#
-#     At least one of scalar or vector attributes must be provided.
-#
-#     Attributes:
-#         basepath (pathlib.Path): Path to the parent directory holding the data file
-#         filename (str): Name of the data file
-#         meta (dict): A dictionary of the necessary metadata to load the image
-#         filepath: Joint file path derived from basename and filename (i.e. basename / filename)
-#         image: The 3D image (e.g. binary image, pressure field, etc.)
-#         vector (list[np.ndarray, np.ndarray, np.ndarray]): A list containing 3 np.ndarrays of the vector components
-#         shape (tuple): The shape of the 3D ndarray containing the image
-#         nx (int): Width of the image (in voxels)
-#         ny (int): Height of the image (in voxels)
-#         nz (int): Number of slices of the image
-#         magnitude: Magnitude of vector (if vector is provided)
-#     """
-#     basepath: pathlib.Path
-#     filename: str
-#     meta: field(default_factory=dict) = None
-#     filepath: str = field(init=False)
-#     image: np.ndarray = field(init=False)
-#
-#     def __post_init__(self):
-#         self.filepath = os.path.join(self.basepath, self.filename)
-#         self.basename, self.ext = self.filename.rsplit('.', 1)
-#         self.image = read_image(self.filepath, meta=self.meta)
-#
-#         # Add 3rd axis if image is 2D
-#         if self.image.ndim == 2:
-#             self.image = self.image[np.newaxis, :, :]
-#
-#         self.nz, self.nx, self.ny = self.image.shape
-#
-#         # TODO add multiple fields (ex. velocity field)
-#         # TODO add functionality for coordinate data
-
-
-# TODO combine VectorImage and Image classes
-# @dataclass
-# class Vector(Image):
-#     scalar: np.ndarray = None
-#     vector: list = None
-#
-#     def __post_init__(self):
-#         assert self.scalar is not None or self.vector is not None, "Provide either scalar or vector data"
-#
-#         if self.scalar.ndim == 2:
-#             self.scalar = self.scalar[np.newaxis, :, :]
-#
-#         self.nz, self.nx, self.ny = self.scalar.shape
-#
-#
-#         self.magnitude = np.sqrt(self.vector[0]**2 + self.vector[1]**2 + self.vector[2]**2)
-
-if __name__ == "__main__":
-    scalar_path = pathlib.Path('C:/Users/bcc2459/Documents/dpm_tools/data/3_fractures.tif')
-    img = Image(filepaths=[scalar_path, scalar_path, scalar_path, scalar_path], meta=[{}, {}, {}, {}])
-    print(img.scalar)
-

dpm_tools/metrics/__init__.py

@@ -25,11 +25,11 @@
     minkowski_map
 """
 
-from ._maps import slicewise_edt, slicewise_mis, edt, sdt, mis, chords, time_of_flight, constriction_factor, minkowski_map
+from .maps import slicewise_edt, slicewise_mis, edt, sdt, mis, chords, time_of_flight, constriction_factor, minkowski_map
 
-from ._feature_utils import _morph_drain_config, _set_linear_trend
+from .feature_utils import _morph_drain_config, _set_linear_trend
 
-from ._scalars import minkowski_functionals, morph_drain, heterogeneity_curve
+from .scalars import minkowski_functionals, morph_drain, heterogeneity_curve
 
 # from .binary_configs import *
 

dpm_tools/metrics/_minkowski_utils.py

@@ -40,17 +40,11 @@ def initialize_mapping(n_dim):
 
 @jit('u1[:, :](u1[:, :], u8, u8)', nopython=True, parallel=False)
 def get_binary_configs_2d(image, dim0, dim1):
-    # n_threads = numba.get_num_threads()
-    # chunk_size = n_threads * 8
-    # n_chunks = (dim0 // (chunk_size)) + 1
 
     mask = np.zeros((dim0 - 1, dim1 - 1), dtype=np.uint8)
 
     IC = initialize_mapping(2)
 
-    # for chunk in prange(n_chunks):
-    #     start_x = chunk * chunk_size
-    #     end_x = min(start_x + chunk_size + 2, dim0)
     for x in range(dim0 - 1):
         for y in range(dim1 - 1):
             mask_val = (int(image[x, y] == 1) +

dpm_tools/metrics/_maps.py → dpm_tools/metrics/maps.py

@@ -3,7 +3,7 @@
 from edt import edt as edist
 
 from ._minkowski_coeff import *
-from ._feature_utils import pad_to_size, create_kernel, _centered
+from .feature_utils import pad_to_size, create_kernel, _centered
 from ._fft_backends import _get_backend
 from ._minkowski_utils import *
 
@@ -95,7 +95,6 @@ def slicewise_mis(image, **kwargs) -> np.ndarray:
     Returns:
         numpy.ndarray: Maximum inscribed sphere computed on each slice (maximum inscribed disk)
     """
-    # ? why do we pad this?
     input_image = np.pad(array=image.copy(), pad_width=((0, 0), (0, 0), (0, 1)), constant_values=1)
 
     # Calculate slice-wise local thickness from PoreSpy
@@ -122,7 +121,6 @@ def chords(image) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
     sz_y = np.zeros_like(ellipse_area)
     for i in range(image.shape[0]):
         # Calculate the chords in x and y for each slice in z
-        # chords = ps.filters.apply_chords_3D(image)
         chords_x = ps.filters.apply_chords(im=image[i, :, :], spacing=0, trim_edges=False, axis=0)
         chords_y = ps.filters.apply_chords(im=image[i, :, :], spacing=0, trim_edges=False, axis=1)
 

dpm_tools/metrics/_scalars.py → dpm_tools/metrics/scalars.py

@@ -2,7 +2,7 @@
 from typing import Tuple
 import edt
 from ._minkowski_coeff import contributions_2d, contributions_3d
-from ._feature_utils import _morph_drain_config, _get_heterogeneity_centers_3d
+from .feature_utils import _morph_drain_config, _get_heterogeneity_centers_3d
 from ._minkowski_utils import get_configs_histogram_2d, get_configs_histogram_3d
 
 
@@ -61,8 +61,6 @@ def _minkowski_3d(image: np.ndarray) -> Tuple[float, float, float, float]:
 
     # Get the isotropic configurations (3D)
     nx, ny, nz = image.shape
-    # bin_img = np.transpose(binary_image, (2, 1, 0))
-    # bin_img = np.transpose(binary_image, (1, 2, 0))
     configs_hist = get_configs_histogram_3d(image, nx, ny, nz)
     v3 = np.sum(contributions_3d["v3"] / 8. * configs_hist)
     v2 = np.sum(contributions_3d["v2"] / 24. * 4 * configs_hist)

dpm_tools/segmentation/segment.py

@@ -26,7 +26,7 @@ def statistical_region_merging(image: np.ndarray, Q: float = 5.0, normalize: boo
         numpy.ndarray: A numpy.ndarray labeled image of the same shape and datatype as the input image.
     """
     # Get image dimensions
-    image_dims = str(image.ndims)
+    image_dims = str(image.ndim)
     assert image_dims in ["2", "3"], "Image must be either 2 or 3 dimensional ndarray"
 
     # Get image datatype
@@ -40,6 +40,7 @@ def statistical_region_merging(image: np.ndarray, Q: float = 5.0, normalize: boo
         image[image < img_min] = img_min
         image[image > img_max] = img_max
         image = (image - img_min) / (img_max - img_min) * np.iinfo(image.dtype).max
+        image = image.astype(image_dtype)
 
     func_dict = {"2":
                     {
@@ -55,21 +56,22 @@ def statistical_region_merging(image: np.ndarray, Q: float = 5.0, normalize: boo
                     }
                 }
 
-    srg_obj = func_dict[image_dims][image_dtype](image, Q)
-    srg_obj.segment()
 
-    return srg_obj.get_result()
+    srm_obj = func_dict[image_dims][image_dtype](image, Q)
+    srm_obj.segment()
+    segmentation = srm_obj.get_result()
+
+    return segmentation
 
 
-def seeded_region_growing(image: np.ndarray, seed_image: np.ndarray, normalize: bool = True)
+def seeded_region_growing(image: np.ndarray, seed_image: np.ndarray, normalize: bool = True):
     """ Perform seeded region growing on a gray level image using predefined seeds
     """
-    # TODO: Write wrapper code for dpm_srg
-    image_dims = image.ndims
-    assert image_dims in [2, 3], "Image must be either 2 or 3 dimensional ndarray"
+    image_dims = str(image.ndim)
+    assert image_dims in ["2", "3"], "Image must be either 2 or 3 dimensional ndarray"
 
-    seed_dims = seed_image.ndims
-    assert seed_dims == image_dims, "Seed image must have the same dimensions as the input image"
+    seed_dims = seed_image.ndim
+    assert seed_dims == int(image_dims), "Seed image must have the same dimensions as the input image"
 
     # Get image datatype
     image_dtype = str(image.dtype)
@@ -83,6 +85,7 @@ def seeded_region_growing(image: np.ndarray, seed_image: np.ndarray, normalize:
         image[image < img_min] = img_min
         image[image > img_max] = img_max
         image = (image - img_min) / (img_max - img_min) * np.iinfo(image.dtype).max
+        image = image.astype(image_dtype)
 
     func_dict = {"2":
                     {
@@ -98,7 +101,7 @@ def seeded_region_growing(image: np.ndarray, seed_image: np.ndarray, normalize:
                     }
                 }
 
-    srg_obj = func_dict[image_dims][image_dtype](image, Q)
+    srg_obj = func_dict[image_dims][image_dtype](image, seed_image)
     srg_obj.segment()
 
     return srg_obj.get_result()

dpm_tools/visualization/_3d_vis_utils.py

@@ -50,12 +50,11 @@ def _custom_cmap(vector, color_map: str = 'turbo') -> Tuple[ListedColormap, floa
 
     min_magnitude = np.percentile(log_mag, 25)
     max_magnitude = np.percentile(log_mag, 99)
-    # print(f'Log min. = {min_magnitude}, Log max. = {max_magnitude}')
 
     cmap_modified = cm.get_cmap(color_map, 65535)
     spacing = lambda x: np.log10(x)
     new_cmap = ListedColormap(cmap_modified(spacing(np.linspace(1, 10, 65535))))
-    # return min_magnitude, max_magnitude
+
     return new_cmap, 10 ** min_magnitude, 10 ** max_magnitude
 
 

dpm_tools/visualization/__init__.py

@@ -25,9 +25,9 @@
     plot_medial_axis
 """
 
-from ._plot_2d import hist, plot_slice, make_thumbnail, make_gif, plot_heterogeneity_curve
+from .plot_2d import hist, plot_slice, make_thumbnail, make_gif, plot_heterogeneity_curve
 
-from ._plot_3d import orthogonal_slices, plot_isosurface, bounding_box, plot_glyph, plot_streamlines, plot_scalar_volume, plot_medial_axis
+from .plot_3d import orthogonal_slices, plot_isosurface, bounding_box, plot_glyph, plot_streamlines, plot_scalar_volume, plot_medial_axis
 
 # from ._vis_utils import *
 

dpm_tools/visualization/_vis_utils.py

@@ -48,8 +48,6 @@ def _scale_image(image_data: np.ndarray, scale_to: type = np.uint8) -> np.ndarra
     :rtype: np.ndarray
     """
 
-    # assert image_data.dtype.type is not scale_to, f"Image data is already of type {scale_to.__name__}"
-
     if 'int' in scale_to.__name__:
         dtype_info = np.iinfo(scale_to)
 

dpm_tools/visualization/_plot_2d.py → dpm_tools/visualization/plot_2d.py

@@ -6,7 +6,7 @@
 from typing import Any
 
 from ._vis_utils import _make_dir, _write_hist_csv, _scale_image
-from ..metrics._feature_utils import _sigmoid
+from ..metrics.feature_utils import _sigmoid
 
 
 # TODO Add fig save decorator
@@ -65,8 +65,8 @@ def hist(data,
     #     save_path = f'{_make_dir("./figures")}'
 
     # TODO add write_csv with proper savepath
-    if write_csv:
-        _write_hist_csv(freq, bins, './figures/histogram_csv.csv')
+    # if write_csv:
+    #     _write_hist_csv(freq, bins, './figures/histogram_csv.csv')
 
     # TODO add savefig?