Moving object detection in dynamic backgrounds for surveillance systems

Abstract

Object detection from a moving image sequence is one of the most challenging research area. Detection of moving objects in a video sequence is the fundamental step but a critical task in information extraction for computer vision applications. The main idea in our paper is to detect the moving objects area smoothly. This paper presents a real time approach to determine objects in the dynamic surveillance systems. It is mainly based on frame difference model and background subtraction model; choose one of the approaches according to the surrounding environment and take the advantages of both FDM and BSM. With this model, a scene containing temporal stationary objects, motionless objects, objects having similar chromatic value of background, illumination change effectively in a surveillance application; can detected smoothly the area of the objects. After determining the moving object area, it is eliminates the noise and finally produce the moving objects area. The experimental results show that our proposed approach can detect the moving objects from video sequence robustly and successfully.