How to remove all pixels that aren’t connected to at least 10 another pixels in python [closed]

Is there a way to remove all pixels that aren’t connected to at least 10 other pixels? I want this

to look something like

.

Answer

Using erosion and dilation to remove the grain works very well for this particular image

import numpy as np
import matplotlib.pyplot as plt
from scipy import ndimage


img = np.dot(plt.imread('https://i.stack.imgur.com/ypBGu.png')[...,:3], [0.299, 0.587, 0.114])

img = 1 - (img > .6)
img = ndimage.binary_erosion(img, structure=np.ones((2,2))).astype(int)
img = ndimage.binary_dilation(img, structure=np.ones((3,3))).astype(int)

labeled_array, num_features = ndimage.label(img)
label, counts = np.unique(labeled_array, return_counts=True)
labeled_array[np.isin(labeled_array, label[counts < 200])] = 0

This gives an array with every character’s island pixels labeled

plt.imshow(labeled_array, cmap='gray')
plt.axis('off');

Output