kmean.py

import cv2
import numpy as np

def main():
    
    w = 160
    h = 120
    
    
    cap = cv2.VideoCapture('testx.avi')
    cap.set(3, w)
    cap.set(4, h)
    
    if cap.isOpened():
        ret, frame = cap.read()
        
    else:
        ret = False

    while ret:
    
        ret, frame = cap.read()
        frame = cv2.resize(frame, (540, 540), interpolation = cv2.INTER_LINEAR) 
        (channel_b, channel_g, channel_r) = cv2.split(frame)
        frame = channel_r
        Z = frame.reshape((-1,3))
        Z = np.float32(Z)
    
        criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)
    
        K=2
        ret, label1, center1 = cv2.kmeans(Z, K, None,
                                      criteria, 10, cv2.KMEANS_RANDOM_CENTERS)
        center1 = np.uint8(center1)
        res1 = center1[label1.flatten()]
        output1 = res1.reshape((frame.shape))
    
        cv2.imshow("Original", frame)
        cv2.imshow("Quantized", output1)
        if cv2.waitKey(1) == 27: # exit on ESC
            break

    cv2.destroyAllWindows()
    cap.release()

#if __name__ == "__main__":
main()