[Python] Why copying the numpy array does not take the effect?

I'm trying to write a simple function which will blend two images, black background and hand image, and move the hand image over the background. I managed to move the hand image over the background, but I don't know why the background does not get cleared when I reassign the original background to the temporary image (it works only when I re-declare background in helper function):

import cv2
import numpy as np
import time
import math

SEQUENCE_LENGTH = 15

def rotate(background, hand):
h,w = hand.shape[:2] #Get hand shape
step = math.floor((background.shape[1]-hand.shape[1])/SEQUENCE_LENGTH) #Calculat>

merged = background[:] #Create temporary array for merging bckg and hand image
#background = np.zeros((480,640,3),dtype='uint8')
for i in range(SEQUENCE_LENGTH):
x_offset = i*step
y_offset = 0

merged[y_offset:y_offset+h, x_offset:x_offset + w] = hand
cv2.imshow("Hand", merged)
cv2.waitKey(100)
merged[:] = background[:] ### Why this does not take effect, unless I uncomm>
### background = np.zeros((480,640,3),dtype='uint8')

def main():
bckg = np.zeros((480,640,3),dtype='uint8') #Generate black background
hand = cv2.imread("/home/hand.jpg") # Get hand image
# Define new width and height
new_width = 100
new_height = 100

resized_hand = cv2.resize(hand, (new_width, new_height)) #Resize hand image
rotate(bckg,resized_hand) #Call rotation function
cv2.waitKey(0)
cv2.destroyAllWindows()

if __name__ == "__main__":
main()


The background gets cleared only when I re-declare background=np.zeros((480,640,3),dtype='uint8'). Otherwise I see the remaining of previous hand image in the background:

No remainings:

Clear background:

It looks like the background assignment does not work, the question is why? I checked the datatypes of background and merged image, and they are the same regardless of whether I use the np.zeros assignment or not.

Continue reading...