ENH: use contrast-limited adaptive histogram interpolation

pyproject.toml

@@ -26,6 +26,7 @@ classifiers = [
 ]
 requires-python = ">=3.10"
 dependencies = [
+    "ahe>=0.1.0",
     "matplotlib>=3.5.0",
     "numpy>=1.22, <3",
     "rlic>=0.2.1",

src/lick/_image_processing.py

@@ -53,43 +53,14 @@ class HistogramEqualizer:
     nbins: int
 
     def process(self, image: FArray2D[F]) -> FArray2D[F]:
-        # adapted from scikit-image
-        """Return image after histogram equalization.
-
-        Parameters
-        ----------
-        image : array
-            Image array.
-
-        Returns
-        -------
-        out : float array
-            Image array after histogram equalization.
-
-        Notes
-        -----
-        This function is adapted from [1]_ with the author's permission.
-
-        References
-        ----------
-        .. [1] http://www.janeriksolem.net/histogram-equalization-with-python-and.html
-        .. [2] https://en.wikipedia.org/wiki/Histogram_equalization
-
-        """
-        import numpy as np
-
-        hist, bin_edges = np.histogram(image.ravel(), bins=self.nbins)
-        bin_centers = (bin_edges[:-1] + bin_edges[1:]) / 2.0
-
-        cdf = hist.cumsum()
-        cdf = cdf / float(cdf[-1])
-
-        cdf = cdf.astype(image.dtype, copy=False)
-        out = np.interp(image.flat, bin_centers, cdf)
-        out = out.reshape(image.shape)
-        # Unfortunately, np.interp currently always promotes to float64, so we
-        # have to cast back to single precision when float32 output is desired
-        return out.astype(image.dtype, copy=False)  # type: ignore[no-any-return]
+        import ahe
+
+        return ahe.equalize_histogram(
+            image,
+            nbins=self.nbins,
+            adaptive_strategy={"kind": "tile-interpolation", "tile-into": 8},
+            max_normalized_bincount=0.01,
+        )
 
 
 class LayeringMode(Enum):