杂波背景下自适应IMM机动目标跟踪算法

Multiple Target Tracking Algorithm Based on Adaptive IMM Algorithm in Clutter

针对杂波背景下多机动目标跟踪问题，提出一种基于时变转移概率交互式多模型（IMM）的模糊数据关联跟踪算法。首先，针对传统IMM算法模型转移概率假设为常数导致模型间过度竞争的问题，基于贝叶斯理论，推导出一种时变模型转移概率IMM算法，增强了优势模型的利用率；其次，针对传统JPDA算法由于聚矩阵拆分而导致的计算组合爆炸问题，利用模糊聚类的方法，直接计算相关波门内候选量测与目标间的关联概率，用概率加权对目标进行状态和协方差的更新。仿真实验表明：算法对不同机动目标的跟踪适应性得到增强，相比传统的JPDA算法，在保证跟踪精度的基础上其时间性能比较优越，是一种较为实用的工程应用算法。

The fuzzy probabilistic data association algorithm based on the interacting multiple model （ IMM ） algorithm with the time-varying transition probability is proposed in this paper to solve the problem of muhiple maneuvering targets tracking in clutter. First, the improved IMM algorithm uses the time-varying transition probability based on Bayes theorem to decrease the excessive competition when the transition probability is constant value in traditional algorithm. Second, to solve the problem of the combina- tion explosion when separating the polymer matrix, the proposed algorithm calculates the association probability on the basis of fuzz- y clustering, which is used as weight to update target＇s state and covariance. Simulation results show that the tracking adaptation to different targets has been enhanced and the real-time performance of the tracking is improved under the premise of tracking accura- cy compared with traditional JPDA algorithm. This is a practical engineering application algorithm.