Tracking of fixed shape moving objects based on modified particle filters

High resolution imaging and its processing is one of the most researched areas now-a-days. Recognition of moving object in a remote scene is a challenging task because of presence of multiple objects present there. Similarly, to track an object after the recognition part is also difficult task because of constant change in not only the coordinates of the target object but also the objects that are cause of occlusion for it. To simplify the task, a constant updating of the algorithms have been carried out in this area pertaining to main two tasks in task. One being is how to reduce the time to recognize the object and other is to efficiently track it, possibly in real time. In this paper, a unified approach is presented which efficiently tracks an object using Particle filter. First step is to identify the coordinates of the object and then a bounding box is placed on it. Then tracking algorithm uses particle filter to efficiently track the moving object in successive frames using probabilistic estimation. The particle filter uses posterior distribution of random variables related to Markov chain for estimation.