Python OpenCV - Background Subtraction

Background Subtraction is one of the major Image Processing tasks. It is used in various Image Processing applications like Image Segmentation, Object Detection, etc. OpenCV provides us 3 types of Background Subtraction algorithms:-

• BackgroundSubtractorMOG
• BackgroundSubtractorMOG2
• BackgroundSubtractorGMG

Normally, we can perform background Subtraction using matrix subtraction, i.e, just subtracting the static frame from the video. But this has a lot of drawbacks. It is a very less efficient algorithm for Background subtraction because it does not update itself. This problem is being handled by the Background Subtraction algorithms provided by OpenCV.

Using BackgroundSubtractorMOG

To use BackgroundSubtractorMOG we can use

cv2.bgsegm.createBackgroundSubtractorMOG()

Then we can apply it using the "apply" method on each frame of the video. Consider the below example for a better understanding of the topic. Example: Python3 1==

import numpy as np
import cv2

cap = cv2.VideoCapture('sample.mp4')

# initializing subtractor 
fgbg = cv2.bgsegm.createBackgroundSubtractorMOG() 

while(1):
    ret, frame = cap.read()       

    # applying on each frame
    fgmask = fgbg.apply(frame)  

    cv2.imshow('frame', fgmask)
    k = cv2.waitKey(30) & 0xff
    if k == 27:
        break

cap.release()
cv2.destroyAllWindows()

Output :

Using BackgroundSubtractorMOG2

In the previous subtractor worked fairly well but in real-world situations, there is also a presence of shadows. In BackgroundSubtractorMOG2, we can also detect shadows and in the output of the following code, it's clearly seen. To apply BackgroundSubtractorMOG2, use

cv2.createBackgroundSubtractorMOG2()

Example: Python3 1==

import numpy as np
import cv2

cap = cv2.VideoCapture('sample.mp4')

# initializing subtractor 
fgbg = cv2.createBackgroundSubtractorMOG2()

while(1):
    ret, frame = cap.read()
 
    # applying on each frame
    fgmask = fgbg.apply(frame)

    cv2.imshow('frame', fgmask)  
    k = cv2.waitKey(30) & 0xff
    if k == 27:
        break

cap.release()
cv2.destroyAllWindows()

Output :

Using BackgroundSubtractorGMG

This algorithm combines statistical background image estimation and per-pixel Bayesian segmentation. It employs a probabilistic foreground segmentation algorithm that identifies possible foreground objects using Bayesian inference. To use BackgroundSubtractorGMG, use

cv2.bgsegm.createBackgroundSubtractorGMG()

Note: We will get a black window during first few frames. Python3 1==

import numpy as np
import cv2

cap = cv2.VideoCapture('sample.mp4')

kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))

# initializing subtractor 
fgbg = cv2.bgsegm.createBackgroundSubtractorGMG()

while(1):
    ret, frame = cap.read()

    # applying on each frame
    fgmask = fgbg.apply(frame)

    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_OPEN, kernel)     

    cv2.imshow('frame', fgmask)
    k = cv2.waitKey(30) & 0xff
    if k == 27:
        break

cap.release()
cv2.destroyAllWindows()

Output :