Utility-Based Camera Node Assignment for Pedestrian Tracking in Visual Sensor Networks

Camera assignment is of great importance for pedestrian tracking in visual sensor networks, since it can not only provide best view for tracking but also save communication resources of the networks. This paper proposes a camera assignment method based on fuzzy logic. First, three criteria related to pedestrian´s size, position, view on the image are designed to evaluate the information of the camera nodes. Second, the scores for these three criteria are fused to the value of suitability by proposed fuzzy inference engine. At last, combing with the latest variation of the criteria scores, the camera nodes with highest score are selected for pedestrian tracking. Our method is validated in visual sensor network including eight camera nodes. The experiment results show the superiority of our method over linear combination approach.