Added browsing function with LUT for gray images

.gitignore

@@ -0,0 +1,6 @@
+__pycache__/nutil.cpython-36.pyc
+NUtil.egg-info/dependency_links.txt
+NUtil.egg-info/PKG-INFO
+NUtil.egg-info/requires.txt
+NUtil.egg-info/SOURCES.txt
+NUtil.egg-info/top_level.txt

nutil.py

@@ -0,0 +1,120 @@
+from PIL import Image
+from ipywidgets import interact
+from IPython.display import display
+import warnings
+import numpy as np
+from numba import njit
+from matplotlib.pyplot import get_cmap
+
+@njit
+def _colorim(x, colors):
+    """Colors a grayscale image using a LUT
+    
+    Parameters
+    ----------
+    x : np.ndarray
+        The image to be colored
+    colors : np.ndarray
+        The colors, ensure colors from 0 ... 255
+    
+    Returns
+    -------
+    np.ndarray (RGBA)
+        The input image colored
+    """
+    im = np.zeros(x.shape+(4,), dtype=np.uint8)
+
+    for i in range(x.shape[0]):
+        for j in range(x.shape[1]):
+            im[i,j] = colors[x[i,j]]
+
+    return im
+
+def browse(x, axis=0, resize=None, enhance_contrast=True, antialiasing=False, cutoff_min=None, cutoff_max=None, cmap=None):
+    """Browsing through an image stack seemlessly
+    
+    Parameters
+    ----------
+    x : np.ndarray
+        An grayscale or RGB image and stack
+    axis : int, optional
+        browsing axis, by default 0
+    resize : int, optional
+        resize image by a factor, e.g. 2, by default None
+    enhance_contrast : bool, optional
+        enhance contrast of image using min/max, by default True
+    antialiasing : bool, optional
+        smooth image when resizing, by default False
+    cutoff_min : int, optional
+        clip minimum value, by default None
+    cutoff_max : int, optional
+        clip maximum value, by default None
+    cmap: str, optional
+        applies colormap to grayscale image, by default None
+    
+    Raises
+    ------
+    warning
+        when float arrays are shown
+    """
+    s = x.shape
+
+    assert type(axis) == int, "axis must be an interger"
+    assert len(s) >= 2, "image must be at least 2D"
+
+    # Can only show single images, no deeper inspection possible
+    if axis > 0 and (len(s) == 2 or (len(s) == 3 and s[-1] == 3)):
+        assert True, print("The array of shape {} cannot be browsed deeper".format(s))
+
+    # Make single image a pseudo-stack
+    if axis == 0 and (len(s) == 2 or (len(s) == 3 and s[-1] == 3)):
+        x = x[None]
+        s = x.shape
+
+    # Normally use nearest neighbour interpolation
+    RESIZE_FLAG = Image.NEAREST
+
+    if antialiasing:
+        RESIZE_FLAG = Image.ANTIALIAS
+
+    if cmap:
+        # Get cmap from matplotlib
+        c = get_cmap(cmap)
+        # Get colors from cmap and scale it from [0,1] to [0, 255]
+        colors = np.array([np.array(c(i))*255 for i in np.linspace(0, 1, 256)], dtype=np.uint8)
+
+    @interact
+    def __(a:(0, s[axis]-1)):
+        # Dynamic fancy indexing using axis as argument
+        im = np.take(x, a, axis)
+
+        # If data should be shown in a specific range
+        if cutoff_min or cutoff_max:
+            im = np.clip(im, cutoff_min, cutoff_max)
+
+        # If data is a float numpy array,
+        # raise warning and convert to uint8
+        if im.dtype == np.float32 or im.dtype == np.float64 or im.dtype == np.float64:
+            warnings.warn("Your dtype is {}, will be converted".format(im.dtype)+\
+                "to uint8 for display purposes")
+            im = im.astype(np.uint8)
+
+        if enhance_contrast:
+            im -= im.min()
+            im = (im.astype(np.float32) / im.max() * 255).astype(np.uint8)
+
+        # If grayscale image should be colored using a LUT
+        if cmap and len(im.shape) == 2:
+            im = _colorim(im, colors)
+
+        # Convert numpy array to PIL Image
+        im_pil = Image.fromarray(im)
+
+        # If image is too small and people asked for resizing
+        if resize is not None:
+            w, h = im.shape[:2]
+            im_pil = im_pil.resize((int(h*resize), 
+                                    int(w*resize)), RESIZE_FLAG)
+
+        # Show image
+        display(im_pil)

setup.py

@@ -0,0 +1,9 @@
+from distutils.core import setup
+from setuptools import find_packages
+
+setup(name='NUtil',
+      version='0.1',
+      author='Andreas Kist',
+      author_email='[email protected]',
+      packages=find_packages(),
+      install_requires=['numpy', 'Pillow', 'numba', 'matplotlib'])