Research On Consensus Based Distributed Target Tracking In Visual Sensor Networks

Target tracking based on wireless sensor networks has recently gained huge popularity in many important applications such as building monitoring and observation,dynamic positioning and tracking,urban management,disaster relief and national defense.It has been a popular research at home and abroad.In the methods of the target tracking based on state estimation,the distributed schemes usually depend on the point-to-point communication between sensor nodes,which have the advantages of high stability,strong fault tolerance and low calculation cost.In a sensor network,the sensor can obtain multiple measurements about the target.The purpose of the distributed schemes is to use all measurements in the network to achieve an accurate estimate of the target state,and there is no central processor.The target tracking technology based on visual network are discussed and the methods for improving the target tracking accuracy is considered in this paper.An Information-weighted Consensus Filtering(ICF)algorithm based on consensus algorithm is proposed,which is also applied to the nonlinear observation model.The major contributions of the paper are as follows:First,based on a lot of literatures,the research significance the target tracking in sensor networks and its research status are reviewed briefly.The advantages and disadvantages of several target tracking methods based on visual sensor networks are summarized.Secondly,two traditional distributed state estimation methods are introduced.Briefly,describe the average consensus algorithm.Then,two kinds of distributed target tracking methods based on Wireless Sensor Networks(WSN)are introduced,namely,the Kalman Consensus Filter(KCF)and the generalized Kalman Consensus Filter(GKCF).Focus on the state estimation theory,the differences between these two algorithms are compared theoretically.Thirdly,considering the characteristics of vision sensor nodes with limited field of view,the presence of na?ve nodes which have no measurement information motivates this paper to propose the ICF algorithm which can be applied in target tracking in camera networks.Specific contents including: the feature of vision networks is first introduced.Then,based on centralized maximum a posteriori estimation the information form of the centralized state estimation of the target isobtained.Next,the distributed implementation of the target state estimation equations is derived based on the consensus algorithm.Finally,combining with the dynamic model of the target state,the ICF algorithm is proposed.The advantages of ICF algorithm compared with other algorithms are analyzed theoretically.The simulation results show that the ICF algorithm has higher accuracy among the traditional KCF algorithm and GKCF algorithm at any given computation and communication resource limit.The tracking performance tends to centralized schemes.Finally,an extended Information-weighted Consensus Filter(EICF)algorithm based on nonlinear observation model is implemented.Specifically,the nonlinear observation model of carame networks is described.Then,according to the extended Kalman filter algorithm and the consensus algorithm,the distributed forms of the target state estimation and the information matrix are obtained based on nonlinear model.Lastly,combining the target state dynamic model,EICF is proposed.Simulation results verify the effectiveness of the proposed algorithm.