Motion Detection using OpenCV in Python

In this tutorial, we will perform Motion Detection using OpenCV in Python. When the Python program detects any motion, it will draw a blue rectangle around the moving object.

Please visit the OpenCV documentation page to know more about the library and all its functions. We will use videos from the webcam on our computer for motion detection using OpenCV in Python.

Let’s begin!

Step by step guide for motion detection in the Python program

Below is the step by step guide for this small Python project:

Import OpenCV and Creating VideoCapture object

Ensure that you have installed OpenCV on your PC. After the installation is complete, import the library.

import cv2

We then need to create a VideoCapture object to read the frames from the input ie. our webcam video. If you want to work with another input file already saved on your PC, you can just type its path instead of the 0.

cap=cv2.VideoCapture(0)

Reading our first frame

The first frame typically means it contains only the background. It the reference frame of our program. If there is any difference in the current frame with respect to the first frame, it means motion is detected. We store our first frame in the frame1 variable.

So, The first line is to read the frame. We then convert the colored frame to B&W since we do not need colors to detect motion. Then we smooth out the image using GaussianBlur.

ret1,frame1= cap.read()
gray1 = cv2.cvtColor(frame1, cv2.COLOR_BGR2GRAY)
gray1 = cv2.GaussianBlur(gray1, (25, 25), 0)
cv2.imshow('window',frame1) 

Reading Subsequent frames

We then write an infinite while loop to read the next frames.

while(True):
    ret2,frame2=cap.read()
    gray2 = cv2.cvtColor(frame2, cv2.COLOR_BGR2GRAY)
    gray2 = cv2.GaussianBlur(gray2, (21, 21), 0)

Now we store the current frame in the frame2 variable and apply the same filters as our first frame. We need a loop since the read() method only captures one frame at a time. So, to capture a continuous video, we have to loop instructions.

Comparing frames

Now we compare our current frame with the first frame, to check if any motion is detected. The absdiff() method gives the absolute value of pixel intensity differences of two frames. The first parameter is the background frame and the second is the current frame.

    deltaframe=cv2.absdiff(gray1,gray2)
    cv2.imshow('delta',deltaframe)

 

Now we have to threshold the deltaframe variable using the cv2.threshold() method. The first parameter is the frame to be thresholded. the second and third are the threshold limits and the last parameter is the method used. The THRESH_BINARY method paints the background in black and motion in white. The dilate() method removes all the gaps in between.

    threshold = cv2.threshold(deltaframe, 25, 255, cv2.THRESH_BINARY)[1]
    threshold = cv2.dilate(threshold,None)
    cv2.imshow('threshold',threshold)

Detecting contours

Using contours, we can find the white images in the black background.

    countour,heirarchy = cv2.findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

We detect contours using the findCountours() method. It returns two variables, contour and hierarchy, and the parameters passed to it are the threshold variable, retrieval method and approximation method.

    for i in countour:
        if cv2.contourArea(i) < 50:
            continue
 
        (x, y, w, h) = cv2.boundingRect(i)
        cv2.rectangle(frame2, (x, y), (x + w, y + h), (255, 0, 0), 2)
    
    cv2.imshow('window',frame2)

We now loop through the contour numpy array and draw a rectangle around the moving object. We get the rectangle bounds using boundingRect() and draw the rectangle onto frame2 using the rectangle() method.

 

And the last lines of code waits for the user to enter a certain character, for instance ‘q’, to break out of the loop and quit all the windows.

    if cv2.waitKey(20) == ord('q'):
      break
cap.release()
cv2.destroyAllWindows()

The output will look like below:

Normal background (first frame)

Comparing frames

Note that the ‘deltaframe’ window and the ‘threshold’ window are all black.

Motion detected

Also, note that only the whites in the ‘threshold’ frame are boxed in the ‘window’ frame.

Complete Python code

import cv2
cap=cv2.VideoCapture(0)

ret1,frame1= cap.read()
gray1 = cv2.cvtColor(frame1, cv2.COLOR_BGR2GRAY)
gray1 = cv2.GaussianBlur(gray1, (21, 21), 0)
cv2.imshow('window',frame1)

while(True):
    ret2,frame2=cap.read()
    gray2 = cv2.cvtColor(frame2, cv2.COLOR_BGR2GRAY)
    gray2 = cv2.GaussianBlur(gray2, (21, 21), 0)
    
    deltaframe=cv2.absdiff(gray1,gray2)
    cv2.imshow('delta',deltaframe)
    threshold = cv2.threshold(deltaframe, 25, 255, cv2.THRESH_BINARY)[1]
    threshold = cv2.dilate(threshold,None)
    cv2.imshow('threshold',threshold)
    countour,heirarchy = cv2.findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    for i in countour:
        if cv2.contourArea(i) < 50:
            continue
 
        (x, y, w, h) = cv2.boundingRect(i)
        cv2.rectangle(frame2, (x, y), (x + w, y + h), (255, 0, 0), 2)
    
    cv2.imshow('window',frame2)
    
    if cv2.waitKey(20) == ord('q'):
      break
cap.release()
cv2.destroyAllWindows()

 

Check out other image processing programs like:

Edge detection using OpenCV in Python