This paper presents moving object detection based on background subtraction for video surveillance system. In all computer vision system, the important step is to separate moving object from background and thus detecting all the objects from video images. The main aim of this paper is to design a bounding box concept and alarm system for the human detection and tracking system in the presence of crowd. The bounding box around each object can track the moving objects in each frame and alarm will fire when object is detected and it can be used to detect crowd or theft and the estimation of crowd. This paper gives the implementation results of bounding box and alarm system for detecting objects and its tracking. The evaluation test performed on proposed system like PSNR, MSE was found.