A Method of Camera Relationship Establishment Based on Temporal and Spatial Information of Video Clips

Camera is widely used in daily life in recent years and causes our life towards the ubiquitous camera environment. The analysis of camera videos offers the functions of detecting, classifying or identifying moving objects. The cooperation of multiple cameras for offering intelligent services has also become an important issue in recent studies. The cooperation relies on a pre-defined camera spatial relationship. It allows the prediction of the next possible camera when an object leaves the FOV (field-of-view) of one camera. To simplify the establishment of the camera spatial relationship, a method is proposed based on the analysis of video clips recorded from multiple cameras. The information of moving objects in the video clips is used to establish correct spatial relationship. The video clips of a community surveillance system with 13 cameras are used for the experiment study of our method. The results show that the method can establish camera spatial relationship correctly.