Merge pull request #158 from siapy/fix

Fix

docs/api/core/exceptions.md

@@ -0,0 +1,2 @@
+
+::: siapy.core.exceptions

mkdocs.yml

@@ -66,6 +66,7 @@ nav:
   - API Documentation:
       - Core:
           - Configs: api/core/configs.md
+          - Exceptions: api/core/exceptions.md
           - Logger: api/core/logger.md
           - Types: api/core/types.md
       - Datasets:

siapy/core/types.py

@@ -19,7 +19,7 @@
 
 SpectralType = sp.io.envi.BilFile | sp.io.envi.BipFile | sp.io.envi.BsqFile
 ImageType = SpectralImage | np.ndarray | Image
-ImageSizeType = int | tuple[int, int]
+ImageSizeType = int | tuple[int, ...]
 ImageDataType = (
     np.uint8
     | np.int16

siapy/entities/images.py

@@ -2,7 +2,7 @@
 import sys
 from dataclasses import dataclass
 from pathlib import Path
-from typing import TYPE_CHECKING, Any, Iterable, Iterator
+from typing import TYPE_CHECKING, Any, Iterable, Iterator, Sequence
 
 import numpy as np
 import spectral as sp
@@ -265,7 +265,9 @@ def to_subarray(self, pixels: "Pixels") -> np.ndarray:
         image_arr_area[v_norm, u_norm, :] = image_arr[pixels.v(), pixels.u(), :]
         return image_arr_area
 
-    def mean(self, axis: int | tuple[int] | None = None) -> float | np.ndarray:
+    def mean(
+        self, axis: int | tuple[int, ...] | Sequence[int] | None = None
+    ) -> float | np.ndarray:
         image_arr = self.to_numpy()
         return np.nanmean(image_arr, axis=axis)
 

siapy/entities/pixels.py

@@ -1,6 +1,6 @@
 from dataclasses import dataclass
 from pathlib import Path
-from typing import Annotated, ClassVar, Iterable, NamedTuple
+from typing import Annotated, ClassVar, Iterable, NamedTuple, Sequence
 
 import numpy as np
 import pandas as pd
@@ -32,6 +32,7 @@ def from_iterable(
                 Annotated[int, "u coordinate on the image"],
                 Annotated[int, "v coordinate on the image"],
             ]
+            | Sequence[int]
         ],
     ) -> "Pixels":
         df = pd.DataFrame(iterable, columns=[Pixels.coords.U, Pixels.coords.V])

siapy/features/features.py

@@ -28,7 +28,15 @@ def __init__(
         feateng_steps: int = 2,
         featsel_runs: int = 5,
         max_gb: int | None = None,
-        transformations: list | tuple = ("1/", "exp", "log", "abs", "sqrt", "^2", "^3"),
+        transformations: list[str] | tuple[str, ...] = (
+            "1/",
+            "exp",
+            "log",
+            "abs",
+            "sqrt",
+            "^2",
+            "^3",
+        ),
         apply_pi_theorem: bool = True,
         always_return_numpy: bool = False,
         n_jobs: int = 1,
@@ -78,7 +86,15 @@ def __init__(
         feateng_steps: int = 2,
         featsel_runs: int = 5,
         max_gb: int | None = None,
-        transformations: list | tuple = ("1/", "exp", "log", "abs", "sqrt", "^2", "^3"),
+        transformations: list[str] | tuple[str, ...] = (
+            "1/",
+            "exp",
+            "log",
+            "abs",
+            "sqrt",
+            "^2",
+            "^3",
+        ),
         apply_pi_theorem: bool = True,
         always_return_numpy: bool = False,
         n_jobs: int = 1,

siapy/features/helpers.py

@@ -16,7 +16,7 @@
 
 class FeatureSelectorConfig(BaseModel):
     k_features: Annotated[
-        int | tuple | str,
+        int | str | tuple[int, ...],
         "can be: 'best' - most extensive, (1, n) - check range of features, n - exact number of features",
     ] = (1, 20)
     cv: int = 3
@@ -38,7 +38,7 @@ def feature_selector_factory(
     problem_type: Literal["regression", "classification"],
     *,
     k_features: Annotated[
-        int | tuple | str,
+        int | str | tuple[int, ...],
         "can be: 'best' - most extensive, (1, n) - check range of features, n - exact number of features",
     ] = (1, 20),
     cv: int = 3,

siapy/transformations/corregistrator.py

@@ -1,3 +1,5 @@
+from typing import Sequence
+
 import matplotlib.pyplot as plt
 import numpy as np
 
@@ -23,10 +25,10 @@ def map_affine_approx_2d(points_ref: np.ndarray, points_mov: np.ndarray) -> np.n
 
 
 def affine_matx_2d(
-    scale: tuple[float, float] = (1, 1),
-    trans: tuple[float, float] = (0, 0),
+    scale: tuple[float, float] | Sequence[float] = (1, 1),
+    trans: tuple[float, float] | Sequence[float] = (0, 0),
     rot: float = 0,
-    shear: tuple[float, float] = (0, 0),
+    shear: tuple[float, float] | Sequence[float] = (0, 0),
 ) -> np.ndarray:
     """Create arbitrary affine transformation matrix"""
     rot = rot * np.pi / 180

siapy/utils/images.py

@@ -6,6 +6,7 @@
 import spectral as sp
 
 from siapy.core import logger
+from siapy.core.exceptions import InvalidInputError
 from siapy.core.types import ImageDataType, ImageType
 from siapy.entities import SpectralImage
 from siapy.transformations.image import rescale
@@ -172,7 +173,7 @@ def convert_radiance_image_to_reflectance(
     panel_correction: np.ndarray,
     save_path: Annotated[
         str | Path | None, "Header file (with '.hdr' extension) name with path."
-    ],
+    ] = None,
     **kwargs: Any,
 ) -> np.ndarray | SpectralImage:
     image_ref_np = image.to_numpy() * panel_correction
@@ -187,14 +188,27 @@ def convert_radiance_image_to_reflectance(
 def calculate_correction_factor_from_panel(
     image: SpectralImage,
     panel_reference_reflectance: float,
-    panel_shape_label: str = "reference_panel",
-) -> np.ndarray | None:
-    panel_shape = image.geometric_shapes.get_by_name(panel_shape_label)
-    if panel_shape is None:
-        return None
-
-    panel_signatures = image.to_signatures(panel_shape.convex_hull())
-    panel_radiance_mean = panel_signatures.signals.mean()
+    panel_shape_label: str | None = None,
+) -> np.ndarray:
+    if panel_shape_label:
+        panel_shape = image.geometric_shapes.get_by_name(panel_shape_label)
+        if not panel_shape:
+            raise InvalidInputError(
+                input_value={"panel_shape_label": panel_shape_label},
+                message="Panel shape label not found.",
+            )
+        panel_signatures = image.to_signatures(panel_shape.convex_hull())
+        panel_radiance_mean = panel_signatures.signals.mean()
+
+    else:
+        temp_mean = image.mean(axis=(0, 1))
+        if not isinstance(temp_mean, np.ndarray):
+            raise InvalidInputError(
+                input_value={"image": image},
+                message=f"Expected image.mean(axis=(0, 1)) to return np.ndarray, but got {type(temp_mean).__name__}.",
+            )
+        panel_radiance_mean = temp_mean
+
     panel_reflectance_mean = np.full(image.bands, panel_reference_reflectance)
     panel_correction = panel_reflectance_mean / panel_radiance_mean
     return panel_correction

tests/utils/test_utils_images.py

@@ -142,7 +142,7 @@ def shape(self) -> tuple[int, int, int]:
         assert save_path.exists()
 
 
-def test_calculate_correction_factor_from_panel(spectral_images):
+def test_calculate_correction_factor_from_panel_with_label(spectral_images):
     pixels = Pixels.from_iterable([(900, 1150), (1050, 1300)])
     rect = Shape.from_shape_type(
         shape_type="rectangle", pixels=pixels, label="reference_panel"
@@ -156,7 +156,6 @@ def test_calculate_correction_factor_from_panel(spectral_images):
         panel_shape_label="reference_panel",
     )
 
-    assert panel_correction is not None
     assert isinstance(panel_correction, np.ndarray)
     assert panel_correction.shape == (image_vnir.bands,)
 
@@ -168,6 +167,18 @@ def test_calculate_correction_factor_from_panel(spectral_images):
     )
 
 
+def test_calculate_correction_factor_from_panel_without_label(spectral_images):
+    image_vnir = spectral_images.vnir
+    panel_correction = calculate_correction_factor_from_panel(
+        image=image_vnir,
+        panel_reference_reflectance=0.3,
+    )
+    direct_panel_calculation = np.full(image_vnir.bands, 0.3) / image_vnir.mean(
+        axis=(0, 1)
+    )
+    assert np.array_equal(direct_panel_calculation, panel_correction)
+
+
 def test_convert_radiance_image_to_reflectance_without_saving(spectral_images):
     image_vnir = spectral_images.vnir
     panel_correction = np.random.default_rng().random(image_vnir.bands)