A new background updating model for motion detection considering future frame
Arifur Rahaman, Md. Mehedi Hasan, Riaz Ahmed, Mirza Mohd Shahriar Maswood, Md. Mostafizur Rahman · 2014
Motion detection is now very much popular for its different type of applications. And the most important thing for motion detection is the background updating procedure. For background updating, all the existing work is done actually depend on the present and past frame. In this paper we proposed a new background updating model for motion detection which mainly differs by considering the future frame. This work is done by taking a number of frames initially to find median from these frames and stored as the past background. Then some frames are taken as future frames and computes a future background from these frames by finding the median. Now if motion occurs in any frame which lies between the past background and this updated future background can be detected easily and accurately by comparing the frames with both of these two updated background. And after some moment the updated future background will be the past background and again some of the frames will be considered as the future frames to calculate the new future background. Our proposed method can update the background accurately as well as can detect motion or moving object perfectly. We have applied this method on various real time data and got promising results.