Background subtraction algorithm for real time video?

How do I perform background subtraction for video input from my web cam? I know how to do this for images and video files, but not for real time video.

[nabihaa izzaty]

Hi David, can I know the code for background substraction for the image and the video files?

[Image Analyst]

nabihaa, my code in my answer gives code to subtract an image (generated from the average of some prior frames) from the current frame. Of course you could adapt it to subtract a constant, fixed image that you read from disk if you want. That should be trivial.

Answer 1

[Image Analyst]

This is done in my demo, attached below. You just need to adapt it to call getsnapshot instead of read() so that it get image from your webcam rather than a video file.

[Pruthvi]

Thank you!

[iup geii amiens]

Hello Sir,

please, can I know you are refering to which method in your demo??

Thank you very much for all what you are doing!

[Image Analyst]

The method is called "subtraction".

[iup geii amiens]

Thank you for your response but there is a lot of methods working on "substraction" or "Background substraction" :)

[iup geii amiens]

Hello Sir, I tried to implement the method of «A Robust Background Subtraction and Shadow Detection» for background subtraction and it is working with fixed thresholds Talpha2 et Talpha2. But i want to make these thresholds automatic, Do you have any idea please? Talpha1 must be the value of alphan at r (detection rate) and Talpha2 the value of alphan at (1-r). Thanks a lot PS; my method is attached below.

[Image Analyst]

lup, I have no idea what those are or how to calculate them. Why can't you just make a function to determine Talpha1 and Talpha2 "automatically" based on other parameters?

[david csernay]

I know this is an old theard, but could somebody pls attach a final version of the code for real time video background subtraction? I get errors when i change read() to getsnapshot(). thank you

[Mrityunjay Jha]

Hello Sir. Can I get the description for this code, so that it will be easy to understand?

[Image Analyst]

I'm attaching the latest version. I think it's pretty well commented. Again, it's for a video but you need to change read() to getsnapshot() or whatever the command for a web cam is if you want to do it in real time rather than from a pre-recorded video.

Answer 2

[Nitin]

You could get the frames from each video and apply your algorithm on each one. Is speed the issue here?