Occlusion Detection with Background Elimination and Moving Object Tracking

Abstract

An occlusion is said to be a situation when few parts of an object cannot be viewed by human vision. Detecting an object is quite difficult to identify and not so easy to track the moving objects during occlusion. It can be achieved in two steps. The first step is to identify the foreground or background image frame by frame, by naming each and every pixel in the frame. The next step is to compare the observations at each point in the sequence of the moving object with occlusion. It can be done by subtracting the background which yields pixel as combination of Gaussians. Then the distribution of Gaussian have to be calculated to conclude the result from the background process. This method is very much useful to identify the motions in surveillance camera, repeated suspicious movement and elongated changes in the scene.