Background modeling using univariate Gaussian density function.

I need to expose background model from 10 consequtive frames, not a video. Also, I need to display both mean and standard deviation images. I got stuck because I could not find any similar project for reference. Thanks in advance.

Answer 1

[yanqi liu]

yes，sir，may be use createBackgroundSubtractorMOG2，such as

import cv2 as cv
import numpy as np

vid = cv.VideoCapture("D:/Program Files/Polyspace/R2019a/toolbox/images/imdata/traffic.avi")
mog = cv.createBackgroundSubtractorMOG2()
se = cv.getStructuringElement(cv.MORPH_RECT, (3, 3))
while True:
    ret, imi = vid.read()
    if ret is True:
        fm = mog.apply(imi)
        ret, bw = cv.threshold(fm, 220, 255, cv.THRESH_BINARY)
        bw = cv.morphologyEx(bw, cv.MORPH_OPEN, se)
        bg = mog.getBackgroundImage()
        cv.imshow("left_bg&right_frame",np.concatenate((bg, imi), axis=1))        
        c = cv.waitKey(50)
    else:
        break

cv.destroyAllWindows()

[Burak Karakus]

Sir, how could I modify this for consecutive frames as input? Should I use a for loop with createBackgroundSubtractorMOG2 instructor? I want to evaluate density value of each pixel using mean and standard deviation parameters via Gaussian density function? Thank you for your interest.

[yanqi liu]

yes，sir，may be upload your video file to make some analysis. this is use python opencv method to process