save normal image as pyramidal tif

wholeslidedata/accessories/asap/convert_image.py

@@ -0,0 +1,67 @@
+import numpy as np
+from PIL import Image
+from pathlib import Path
+
+from wholeslidedata import WholeSlideImage
+from wholeslidedata.accessories.asap.imagewriter import WholeSlideMaskWriter, WholeSlideImageWriter
+
+def take_closest_number(l, number):
+    return min(l, key=lambda x: abs(x - number))
+
+def take_closest_number_index(l, number):
+    closest = take_closest_number(l, number)
+    for ind, val in enumerate(l):
+        if val==closest: return ind
+
+def save_normal_image_as_tif(path, out_path, spacing):
+    """ saves a normal image (png, jpeg) as pyramidal tif """
+    img = Image.open(str(path))
+    arr = np.array(img).squeeze()
+    img.close()
+    save_array_as_image(arr, path=out_path, spacing=spacing)
+
+def save_array_as_image(arr, path, spacing, tile_size=512):
+    """ saves an array as pyramidal tif """
+    if len(arr.shape)==2:
+        arr = arr[:,:,None]
+        writer = WholeSlideMaskWriter()
+    else:
+        writer = WholeSlideImageWriter()
+    shape = arr.shape
+    writer.write(
+        path=str(path), spacing=spacing, dimensions=(shape[1], shape[0]),
+        tile_shape=(tile_size, tile_size),
+    )
+
+    for col in range(0, shape[1]+tile_size, tile_size): #+tile_size if array not divisible by tile_size
+        for row in range(0, shape[0]+tile_size, tile_size):
+            tile = arr[row:row+tile_size, col:col+tile_size]
+            if len(tile)==0: continue #for the edge-case
+            if tile.shape[0]!=tile_size or tile.shape[1]!=tile_size:
+                pad = ((0, tile_size-tile.shape[0]),(0, tile_size-tile.shape[1]),(0,0))
+                tile = np.pad(tile, pad, mode='constant')
+            writer.write_tile(tile=tile, coordinates=(col,row))  #col,row (x,y)
+    writer.save()
+
+
+def save_normal_image_as_mask(path, slide_path, out_path):
+    """ saves an image-mask (png, jpeg) as pyramidal tif for a given slide """
+    reader = WholeSlideImage(str(slide_path), backend='asap')
+    print(zip(reader.spacings, reader.shapes))
+    spacings = reader.spacings
+    shapes = reader.shapes
+    reader.close()
+
+    mask = Image.open(str(path))
+    mask_arr = np.array(mask)
+    h,w = mask.height, mask.width
+    mask.close()
+
+    heights = [hw[0] for hw in shapes]
+    level = take_closest_number_index(heights, h)
+    print('for mask shape(%d,%d) closest level %d with spacing  %.3f, shape %s' % \
+          (h, w, level, spacings[level], str(shapes[level])))
+
+    spacing = spacings[level]
+    return save_array_as_image(mask_arr, spacing=spacing, path=out_path)
+