本文把Ｂｏｕｎｄｉｎｇ估计引入目标跟踪领域，提出一种用状态估计集几何中心作为目标滤波值的递推算法。该算法使用椭圆集来描述状态的不确定性并给和系统和观测噪声的界，以最小值椭圆作为更新准则。仿真结果给出了本算法与Ｋａｌｍａｎ滤波算法的性能比较以及界内噪声分布不同时跟踪性能的差别。

State bounding is applied to target tracking in this paper. A recursive algorithmtising the center of the feasible state estimation set as the filter value of the target is introduced.The algorithm uses ellipsoidal sets to describe the state uncertainties and bound the process andobservation noises, minimum-volume as criteria. Simulation results show the performancecomparison between Kalman filter and this algorithm, and difference while noise distribution inbound varies.

TN95