Mobile Object Tracking in a Video using Kalman Filter

Abstract

Object tracking in a video is the problem of estimating the positions and other related information regarding moving objects in video. Object tracking is a very important task in the field of security automation surveillance systems. For detecting and tracking the moving objects, surveillance system are used. First stage of the system is detecting the moving objects in the video. Second stage of the system is tracking the detected object. Here, detection of the moving object is done by using a simple background subtraction and tracking of moving objects is done by using Kalman filter. The algorithm is applied successfully on standard video datasets. The videos used here for testing have been taken at indoor as well as outdoor environment having moderate to complex environments. Kalman filter tracks an object by assuming the initial state and estimating noise covariance. It provides an efficient method for calculating the state estimation process. An experimental result which came from different moving object video samples shows a very good result. This filter is intended to be robust without being programmed with all environment specific rules