standardize setup and add package structure

README.md

@@ -1,2 +1,8 @@
 # fluidflower_benchmark_analysis
 Image analysis of FluidFlower images for the International Benchmark Study
+
+In order to run the analysis, run:
+python setup.py develop
+
+This install the package benchmark (tentative name) and allows to use
+standardized setups of geometries, analysis tools etc.

rigs/largefluidflower.py → benchmark/rigs/largefluidflower.py

@@ -10,7 +10,7 @@
 from scipy.interpolate import RBFInterpolator
 
 
-class LargeFluidFlower:
+class LargeFluidFlower(daria.AnalysisBase):
     def __init__(
         self,
         baseline: Union[str, Path, list[str], list[Path]],
@@ -28,7 +28,7 @@ def __init__(
             update_setup (bool): flag controlling whether cache in setup
                 routines is emptied.
         """
-        super().__init__(baseline, config, update_setup)
+        daria.AnalysisBase.__init__(self, baseline, config, update_setup)
 
         # Segment the baseline image; identidy water and esf layer.
         self._segment_geometry()
@@ -75,7 +75,7 @@ def _determine_effective_volumes(self) -> None:
         # The setup takes approx. 30 seconds. So the cost is bareable when not switiching
         # between different boxes too often.
 
-        if not Path("volumes.npy").exists():
+        if not Path("tmp/volumes.npy").exists():
             # Determine number of voxels in each dimension - assume 2d image
             Ny, Nx = self.base.img.shape[:2]
             Nz = 1
@@ -274,7 +274,7 @@ def _determine_effective_volumes(self) -> None:
             # Compute effective volume per porous voxel
             self.effective_volumes = porosity * width * height * depth / (Nx * Ny * Nz)
 
-            np.save("volumes.npy", self.effective_volumes)
+            np.save("tmp/volumes.npy", self.effective_volumes)
 
         else:
-            self.effective_volumes = np.load("volumes.npy")
+            self.effective_volumes = np.load("tmp/volumes.npy")

benchmark/utils/fluidflower.py

@@ -0,0 +1,150 @@
+"""
+Module containing the setup for the fluidflower rig, used for the PoroTwin1 optimal control project.
+
+Applicable for uncorrected images (of C1).
+"""
+from pathlib import Path
+from typing import Union
+
+import cv2
+import daria
+import numpy as np
+import skimage
+
+from benchmark.rigs.largefluidflower import LargeFluidFlower
+
+
+# Define specific concentration analysis class to detect mobile CO2
+class CO2MaskAnalysis(daria.BinaryConcentrationAnalysis):
+    """
+    Binary concentration analysis based on a multichromatic HSV comparison
+    and further analysis on the third component, i.e., V, identifying signal
+    strength.
+    """
+
+    def __init__(self, base, color, esf, **kwargs) -> None:
+        super().__init__(base, color, **kwargs)
+
+        # Heterogeneous thresholding
+        self.threshold_value = np.zeros(self.base.img.shape[:2], dtype=float)
+        self.threshold_value[esf] = kwargs.pop("threshold value esf")
+        self.threshold_value[~esf] = kwargs.pop("threshold value non-esf")
+
+        # Fetch parameters for HUE based thresholding
+        self.hue_low_threshold = kwargs.pop("threshold min hue", 0.0)
+        self.hue_high_threshold = kwargs.pop("threshold max hue", 360)
+
+    def _extract_scalar_information(self, img: daria.Image) -> None:
+        """Transform to HSV.
+
+        Args:
+            img (daria.Image): Input image which shall be modified.
+        """
+        img.img = cv2.cvtColor(img.img.astype(np.float32), cv2.COLOR_RGB2HSV)
+
+    def _extract_scalar_information_after(self, img: np.ndarray) -> np.ndarray:
+        """Return 3rd component of HSV (value), identifying signal strength.
+
+        Args:
+            img (np.ndarray): trichromatic image in HSV color space.
+
+        Returns:
+            np.ndarray: monochromatic image
+        """
+
+        # Clip values in hue - from calibration.
+        h_img = img[:, :, 0]
+
+        mask = skimage.filters.apply_hysteresis_threshold(
+            h_img, self.hue_low_threshold, self.hue_high_threshold
+        )
+
+        # Restrict to co2 mask
+        img[~mask] = 0
+
+        # Consider Value (3rd component from HSV) to detect signal strength.
+        return img[:, :, 2]
+
+
+class BenchmarkCO2Analysis(LargeFluidFlower, daria.CO2Analysis):
+    """
+    Class for managing the FluidFlower benchmark analysis.
+    """
+
+    def __init__(
+        self,
+        baseline: Union[str, Path, list[str], list[Path]],
+        config: Union[str, Path],
+        update_setup: bool = False,
+    ) -> None:
+        """
+        Constructor for Benchmark rig.
+
+        Sets up fixed config file required for preprocessing.
+
+        Args:
+            base (str, Path or list of such): baseline images, used to
+                set up analysis tools and cleaning tools
+            config (str or Path): path to config dict
+            update_setup (bool): flag controlling whether cache in setup
+                routines is emptied.
+        """
+        LargeFluidFlower.__init__(self, baseline, config, update_setup)
+        daria.CO2Analysis.__init__(self, baseline, config, update_setup)
+
+    # ! ---- Analysis tools for detecting the different CO2 phases
+
+    def define_co2_analysis(self) -> daria.BinaryConcentrationAnalysis:
+        """
+        Identify CO2 using a heterogeneous HSV thresholding scheme.
+        """
+        co2_analysis = CO2MaskAnalysis(
+            self.base, color="", esf=self.esf, **self.config["co2"]
+        )
+
+        return co2_analysis
+
+    def define_co2_gas_analysis(self) -> daria.BinaryConcentrationAnalysis:
+        """
+        Identify CO2(g) using a thresholding scheme on the blue color channel.
+        """
+        co2_gas_analysis = daria.BinaryConcentrationAnalysis(
+            self.base, color="blue", **self.config["co2(g)"]
+        )
+
+        return co2_gas_analysis
+
+    # ! ----- Analysis tools
+
+    def determine_co2_mask(self) -> daria.Image:
+        """Determine CO2.
+
+        Returns:
+            daria.Image: boolean image detecting CO2.
+        """
+        # Extract co2 from analysis
+        co2 = super().determine_co2()
+
+        # Add expert knowledge. Turn of any signal in the water zone
+        co2.img[self.water] = 0
+
+        return co2
+
+    def determine_co2_gas_mask(self, co2: daria.Image) -> daria.Image:
+        """Determine CO2.
+
+        Args:
+            co2 (daria.Image): boolean image detecting all co2.
+
+        Returns:
+            daria.Image: boolean image detecting CO2(g).
+        """
+        # Extract co2 from analysis
+        co2_gas = super().determine_co2_gas()
+
+        # Add expert knowledge. Turn of any signal outside the presence of co2.
+        # And turn off any signal in the ESF layer.
+        co2_gas.img[~co2.img] = 0
+        co2_gas.img[self.esf] = 0
+
+        return co2_gas

setup.py

@@ -0,0 +1,3 @@
+from setuptools import setup, find_packages
+
+setup(name="benchmark", packages=find_packages())