Cluster non-zero values in a 2D NumPy array

Question

I want to cluster non-zero locations in a NumPy 2D array for MSER detection. Then I want to find the number of points in each cluster and remove those clusters which do not have number of points between some x and y (10 and 300).

I have tried clustering them by searching with neighbouring points but the method fails for concave-shaped non-zero clusters.

[[0, 1, 0, 0, 1],
 [0, 1, 1, 1, 1],
 [0, 0, 0, 0, 0],
 [1, 1, 0, 1, 1],
 [1, 0, 0, 1, 1]]

should output, for x=4 and y=5 (both included)

[[0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0],
 [0, 0, 0, 1, 1],
 [0, 0, 0, 1, 1]]

Answer 1

I'm not sure I have understood your question correctly, but I think scikit-image's label and regionprops could get the job done.

In [6]: import numpy as np

In [7]: from skimage import measure, regionprops

In [8]: img = np.array([[0, 7, 0, 0, 7],
   ...:                 [0, 9, 1, 1, 4], 
   ...:                 [0, 0, 0, 0, 0], 
   ...:                 [2, 1, 0, 2, 1],
   ...:                 [1, 0, 0, 6, 4]])
   ...: 

In [9]: arr = measure.label(img > 0)

In [10]: arr
Out[10]: 
array([[0, 1, 0, 0, 1],
       [0, 1, 1, 1, 1],
       [0, 0, 0, 0, 0],
       [2, 2, 0, 3, 3],
       [2, 0, 0, 3, 3]])

In [11]: print('Label\t# pixels')
    ...: for region in measure.regionprops(arr):
    ...:     print(f"{region['label']}\t{region['area']}")
    ...:
Label   # pixels
1       6
2       3
3       4