how to extract and update background from video sequence

I have a video about traffic scene. now, I want to calculate the percentage of vehicle area on the road area( or the percentage of area foreground on the area background). and the first step for this work is background extraction. I have read many document and Scientific articles about it, one of them recommend me use the mean filter following this formula.

this is link of that article: Scientific article and the results of them are very good, it is exactlly what I want.

I follow his formula and I try to write my code. but It not work! who can help me of give me some advices. this is my code:

vob = VideoReader('NKKN.avi');
frame = vob.read(inf);
vidHeight = vob.Height;
vidWidth = vob.Width;
nFrames = vob.NumberOfFrames;
%%%First-iteration background frame
background_frame = double(frame*0);
redbackground_frame = background_frame(:,:,1);
greenbackground_frame = background_frame(:,:,2);
bluebackground_frame = background_frame(:,:,3);
%calculate background
i = 0;
for k = 1:10 %get background from 10 frame (J=10)
    thisframe = double(read(vob, k));
    %background_frame = background_frame + thisframe;
    redbackground_frame = redbackground_frame + thisframe(:,:,1);
    greenbackground_frame = greenbackground_frame + thisframe(:,:,2);
    bluebackground_frame = bluebackground_frame + thisframe(:,:,3);
    i=i+1;
    disp(i);
end

A = redbackground_frame/i;
B = greenbackground_frame/i;
C = bluebackground_frame/i;

this is my video sequence: video_object

[chfakht chfakht]

Hi, Have you solved the issue ?

[SAASTN]

This has an answer here .

Answer 1

[Image Analyst]

I do something similar to this in my attached demo. But my demo is a weighted average of prior frames instead of calculated in advance. Let me know if you have any questions.

[Leox91]

Which line? I can't find any weighted average.

[Walter Roberson]

The weighting appears to be the line

Background = (1-alpha)* thisFrame + alpha * Background;

[Sritam Patnaik]

Changing alpha to

alpha = 1 / (frame);

and Background to

Background = (1-alpha)* Background + alpha * thisFrame;

in @Image Analyst code did it for me.

[Image Analyst]

That would reduce the effect of the frames in a non-linear manner and would cause the background to not consider the previous frames much at all as you get very far into the video.

[Malik Zulqarnain]

how can i get the background

[Image Analyst]

What do you consider to be the background? Describe it to me in words, like it's frame #1 or whatever.

[Image Analyst]

Is there anyway you can find a frame with no object in it? One way is to just take the mode value of each pixel. This assumes that when the background is displayed, it will have that mode gray level and when something else is there, it won't have that value. So the most common gray level for that pixel should be the value of the background alone.

[Malik Zulqarnain]

Your code for ExtracingMovieFrame.m provide me background with too much color distortion. I have attach the .jpg image of result how can i improve the results

[Image Analyst]

What code did you use? The code I gave hai tran, or the algorithm I suggested for using the mode value? And is your camera still or moving?

[Malik Zulqarnain]

i uses the code that you given above ExtractMoviesFrame.m(i have attached the file) and vedio i uses is taken by stationary camera

[Image Analyst]

Well maybe the algorithm that used for determining the background is not what you want. Try the mode method, if you have enough memory. Or, if you have the Computer Vision System Toolbox, try some of their examples that do that: https://www.mathworks.com/products/computer-vision/code-examples.html