Wireless Camera Sensor Networks Target Tracking Study

Traditional Wireless Sensor Networks(WSN) can only perceive simple scalardata, such as temperature, humidity and light, which provides less information andsimple processing so that WSNs are not available for complex and changableenvironmental monitoring. Based on the demands above, Wireless Camera SensorNetworks(WCSN) emerges as required. With embedded camera as its sensor node,sensor nodes in WCSNs can obtain more intuitive objective physical data as imageand vedio. WCSNs are mostly applied to visual environmental monitoring andextensively used in military, civil and commercial fields due to its convenience,intuition, and rich information, and is a major trend for Wireless Sensor Networks.Target tracking is an important branch in the research for Wireless SensorNetwork, as sensor networks can be randomly deployed and adopt wirelesscommunication, possess self-organizing and robustness abilities, and is extremelyappropriate for moving target tracking. However, traditional WSNs have certainlimitations in target tracking due to its simple scalar data, less data information, poorapplication flexibility, while WCSNs are more suitable for target tracking because itcan not only acquire the coordinates of target through parameters calibration toimplement more direct and preciser target localization, but also obtain targetappearance, so the target tracking can be realized with richer application.This paper first analyzes conventional target tracking algorithms which mostlyuse Kalman Filtering for prediction in the process of target tracking and have hightracking precision but also high computation and time complexity, which goes againstthe need for timeliness. Based on the disadvantages stated above, this paper proposesMoving Trend Dynamic Clustering(MTDC) target tracking algorithm. MTDCimplements target tracking by forming a dynamic cluster which includes a ClusterHead (CH) and several Cluster Members, and the Cluster Head makes dynamicscheduling to select cluster members. MTDC mainly consists of two stages: First,Cluster Head builds a set of neighbour nodes within its communication range, selectssensors which could detect targets as its cluster members; When the target is movingtimely, cluster head updates its cluster members at regular intervals, removes thenodes which lose ability to monitor the target and adds new nodes into cluster; Then,new CH will be elected when the current CH is about to lose the detection of the target; According to the moving direction of the target, cluster member that has thebest moving trend with the target will be chosen as new cluster head, which couldensure the new head cluster has longer tracking period and better tracking precisionand avoid frequent cluster head electing. The cluster is updated timely to track thetarget till it moves outside the monitoring area. The experimental results show thatMTDC has certain precision in target tracking and more efficient compared to otheralgorithms.