mirror of
https://github.com/opencv/opencv.git
synced 2024-12-21 05:28:01 +08:00
51 lines
1.2 KiB
Python
Executable File
51 lines
1.2 KiB
Python
Executable File
#!/usr/bin/env python
|
|
|
|
'''
|
|
K-means clusterization sample.
|
|
Usage:
|
|
kmeans.py
|
|
|
|
Keyboard shortcuts:
|
|
ESC - exit
|
|
space - generate new distribution
|
|
'''
|
|
|
|
# Python 2/3 compatibility
|
|
from __future__ import print_function
|
|
|
|
import numpy as np
|
|
import cv2
|
|
|
|
from gaussian_mix import make_gaussians
|
|
|
|
if __name__ == '__main__':
|
|
cluster_n = 5
|
|
img_size = 512
|
|
|
|
print(__doc__)
|
|
|
|
# generating bright palette
|
|
colors = np.zeros((1, cluster_n, 3), np.uint8)
|
|
colors[0,:] = 255
|
|
colors[0,:,0] = np.arange(0, 180, 180.0/cluster_n)
|
|
colors = cv2.cvtColor(colors, cv2.COLOR_HSV2BGR)[0]
|
|
|
|
while True:
|
|
print('sampling distributions...')
|
|
points, _ = make_gaussians(cluster_n, img_size)
|
|
|
|
term_crit = (cv2.TERM_CRITERIA_EPS, 30, 0.1)
|
|
ret, labels, centers = cv2.kmeans(points, cluster_n, None, term_crit, 10, 0)
|
|
|
|
img = np.zeros((img_size, img_size, 3), np.uint8)
|
|
for (x, y), label in zip(np.int32(points), labels.ravel()):
|
|
c = list(map(int, colors[label]))
|
|
|
|
cv2.circle(img, (x, y), 1, c, -1)
|
|
|
|
cv2.imshow('gaussian mixture', img)
|
|
ch = cv2.waitKey(0)
|
|
if ch == 27:
|
|
break
|
|
cv2.destroyAllWindows()
|