Research On Filter And Correlated Noise Decoupling Strategy Based On Moving Target Tracking

Moving object tracking combined with the state prediction of the moving target and sensormeasurement results is a research direction which related to many fields, including sensormeasurement, computer technology, automatic control and network communication etc.Nowadays, this research direction has instructive meaning both for theoretical study andengineering application in military, industry and civilian. With development of technology,research tasks of moving target tracking become more and more complex，such as the charactersof nonlinear, multi-dimensions and maneuvering model what the targets have; the background ofnon-gaussian. This paper aims at researching target tracking algorithm in the background ofcolored noise and correlated noise, main works is as follows：1.After the study of basic theories and summarizing references, making a brief description ofthe optimal estimation theory, the classification of dynamic system model, and typical filter design,as well as the Dynamic and The Current Situation and Development of object tracking researchdirection. The classification of the dynamic system model and building, typical linear andnonlinear filter is designed, and the dynamic target tracking direction of research and developmentsituation.2.A filtering algorithm is proposed for the environment of colored measurement noise， thealgorithm is based on Kalman filter algorithm，but estimated the colored noise and the covarianceof noise and state straightly, and add the covariance to the calculation of the kalman gain, in orderto amend the noise and system state at the same time. By using the forming filter, and then updatethe estimation with measurements. The algorithm can effectively eliminating the effects of coloredmeasurement noise to filtering result and improve tracking accuracy.3.On the basic of building a correlated noise system model for correlated noise environment,this paper propose a decoupling strategy based on system reconfiguration, this strategy firstly,introduced a variable to system equation for a equivalent transformation, then decoupling thesystem and measurement noise by giving appropriate value to the variable. After that this paperproposed a new filter algorithm on the base of UKF, this algorithm can decoupling the noise in the correlated noise environment, and get a higher filtering accuracy. finished object tracking whenthe noise is correlated.4. Summarize the positive results of this paper, make a prospects of this research direction, atlast, there is a analyzed briefly of problems that could to be improved in this paper and furtherdevelopment could be make of related research.