how can i apply(visi​on.Foregro​undDetecto​r to obtain a foreground model at each input frame of video?

i want to model a background at each input frame of video, i don't know how to do it,any idea please?

Answer 1

[Image Analyst]

I believe Gaussian Mixture Models are usually/often used for background estimation in many videos. Sorry I don't have any demo code for you but look up examples in the Computer Vision System Toolbox.

Answer 2

[Dima Lisin]

The step() method of vision.ForegroundDetector takes the video frame and returns the foreground mask, which is a binary image where foreground pixels have the value of 1, and background pixels have the value of 0. See the example on vision.ForegroundDetector doc page.