Person tracking using mean shift with Gray Level Grouping

Tracking is an essential part of Computer vision. Visual tracking involves two processes known as detection and tracking. A template is designed in the detection stage and further comparison is done in tracking stage. Detection requires extraction of properties of the desired frame and then storing it as a template for further use of it in tracking. During the studies we have often found that tracking generally is not efficient when detection is not proper. Our focus is on improving the process in detection stages where templates are stored for tracking stages. When Tracking is done, there is reduced number of false positives that makes it an efficient algorithm. This technique involves less computational effort that makes it more reliable in real time environment.