基于输入估计的机动目标跟踪方法

Method for Maneuvering Target Tracking Based on Input Estimation

机动目标跟踪方法有变维滤波、输入估计和交互式多模等方法。由于输入估计法采用常速模型作为状态方程对目标进行滤波跟踪,避免了模型集设计而成为研究的热点,但传统输入估计方法跟踪机动目标时存在机动检测时延较长和估计精度不高的缺点。针对这些问题,对输入估计算法进行了改进。采用变检测窗长度提高机动检测的响应速度,通过对检测窗内信息序列的修正显著提高了估计精度。仿真结果表明所提出的算法比原有算法的跟踪性能有显著提高,且对加速度随时间变化的机动也有一定的跟踪性能。

Maneuvering target tracking algorithm includes variable dimension filtering, input estimates, and interactive mul- tiple model method . Because input estimation method using constant speed filtering model as equations of State for target tracking, avoiding the model design, so research hot spots. But traditional input estimation method for tracking a maneuve- ring target exists shortcomins of maneuvering detection delay and lowly estimate accuracy. To address the above problems, an improvement on input estimation algorithms is made. Taking a variable - length detection Windows increase the response speed of maneuver detection, detection in Windows through the amendments significantly improve the estimation accuracy of sequence information. Simulation results indicate that proposed algorithms has significantly improved tracking performance when comparing with the original algorithm, and mobile has some on the acceleration versus time tracking capabilities.