Background Subtraction Method for Object Detection and Tracking

Video object extraction and its tracking is one of the fundamental tasks of computer vision that require a close observation on video content analysis. However, these tasks become sophisticated due to spatial and temporal changes in the video background. In this work, we have proposed a background subtraction algorithm that efficiently localizes the object in the scene. In the next stage, a regional level process is integrated by calculating the Shannon energy and entropy to correctly examine the nonstationary pixels in the frames. In order to extract the object efficiently, the background model is updated to the dynamics changes that reduces the false negative pixels on foreground. Further, an adaptive Kalman filter is integrated to track the object in consecutive frames. Qualitative and quantitative analysis on some experimental videos shows that the method is superior to some existing background subtraction methods used in tracking.