Unattended object detection based on blob tracking

In this study, an effective approach is proposed for detecting unattended objects in visual surveillance systems. In the proposed approach, the regions of foreground are labeled as moving regions which are determined by background subtraction algorithms in a streamed video from a fixed camera. Moving blobs are used to extract information about events occurring in the environment from which the image stream is taken. Each moving region is followed separately to determine whether it forms any permanent region behind. Thus, it can be understood that it is an object left unattended by the tracked moving object. In this paper, the results are shown that obtained with the proposed approach.