Study And Implementation Of Hierarchical Background Modeling Based On PTZ Camera

Because PTZ camera provides more details of scenes and its shooting area isbroader than fix camera, PTZ camera is more and more widely used in publicsupervision systems. As the increasing demand on intelligent supervision systemtoday, many scholars study the object detection and tracking algorithm. Among them,background modeling can effectively improve the tracking methods. However, mostof background modeling methods are only applicable to fix camera and cannot workin PTZ camera which can pan, tilt and zoom.The goal of this paper is to propose a solution of background modeling and targetdetection based on a single PTZ camera.By modifying the texture characteristicsSILTP, SILTP is evolved into CS-SILTP. CS-SILTP is not only more concise but alsohas a better gradient characteristics and practicality. This paper is also extendsCS-SILTP on timeline, and use PKDE to extract foreground. This paper extracts keyframes in different scales of PTZ camera to create a hierarchical background model.Hierarchical background model is able to adapt to a wide shooting range and scaletransformation, so it can be applied to PTZ camera.After the background modelingwith improved CS-SILTP, we need to register the current frame to the key frames ofthe related layer, so that the background model can be corresponding to the currentframe and extract foreground.The experimental results show the extended CS-SILTPgets better effect and stability in the test sets and.Hierarchical background model canbe applied to extract foreground in PTZ camera.