改进卡尔曼滤波的目标跟踪的的研究_杨柳.pdfVIP

27 9 2010 9 : 1006 - 9348( 20 10) 09 - 035 1- 05 杨 柳 ( 1. , 1570 12) : ( EKF) , T aylor , , , , U nscen ed, U n scen ed ( UK F) U n scen ed , S igm a , , Jacob ian H ess ians , , , UK F : ; ; ; : T P 391. 9 : A Study on Target TrackingBased on ImprovedUnscented Transform Kalman Filtering YANG L iu ( D epa rm en of Compu er Sc ience, M udan jiang N orm a l Schoo l, M udan jiang H eilongj iang 1570 12, Ch ina) ABSTRACT: Ex ended K alm an F il er ing is in comm on u sag e in arge rack ing curren ly, bu his m e hod invo lves T aylor se ries expansion, so i has h igh com pu a iona l cos, he ypical va lue and covar iance on ly has f irs de riva ive prec ise, so he prec ision is low er. T o a mi a ra ising he speed and prec ision, U nscen ed ransfo rm has been draw n in- o K a lm an F il e ring in h is pape r, and he m ode l o f UK F has been pub lished. Becau se U nscen ed ransform is based on he heo ry of G au ssian D is ribu ion, so he ypical va lue and covar iance has Th ird- order prec ise hrough Sigm a spo s in his m e hod, he prec ision of f il er ing has been ra ised. I only invo lves he opera ions o f s andard vec o r and m a rix, and has no u se for he ca lcula ion of non- liner m a r ix o f J acob ian and H essians, so he speed o f f il er ing has been increased. Through m ovem en emu la ion, i show s ha for non linear i y arge rack ing sys em, UKF g ives be er accuracy and s ab ili y. KE WORDS: T arge T rack ing; K a lm an F il er ing; G aussian D is r ibu ion; F il er ing prec ision 1 Bucy, Sunahara