Voronoi-Otsu-Labeling on binary images

The Voronoi-Otsu-Labeling algorithm can also be used on binary images, leading to similar results like the binary watershed in ImageJ. It splits roundish objects in case they stick together.

from skimage.io import imread
from napari_segment_blobs_and_things_with_membranes import voronoi_otsu_labeling, threshold_otsu
import stackview

To demonstrate this, we start wit a binary image.

blobs_image = imread("../../data/blobs.tif")
binary_image = threshold_otsu(blobs_image)

binary_image

nsbatwm made image shape (254, 256) dtype int32 size 254.0 kB min 0 max 1

We now apply Voronoi-Otsu-Labeling to the binary image.

label_image = voronoi_otsu_labeling(binary_image, spot_sigma=3.5)

label_image

nsbatwm made image shape (254, 256) dtype int32 size 254.0 kB min 0 max 67

Exercise

Below you see the result of the Watershed in ImageJ. How do you need to modify the label_image result above to turn it again into a binary image like the result of ImageJ?

binary_watershed_imagej = imread("../../data/blobs_otsu_watershed.tif")

stackview.insight(binary_watershed_imagej)

shape (254, 256) dtype uint8 size 63.5 kB min 0 max 255