基于隐马尔可夫模型的线谱跟踪技术

Research on freguency tine tracking based on HMM

线谱检测和跟踪是被动声呐信号处理中的重要内容,本文给出一种基于隐马尔可夫模型的线谱跟踪方法。它采用前向后向算法对LOFAR谱图上线谱进行状态估计,然后根据连续检验对每根线谱的起始和终止时间进行检测,实现对单根线谱和多根线谱的检测与跟踪。通过计算机仿真和海试数据处理,验证了基于隐马尔可夫模型的线谱跟踪技术的有效性和稳定性。

Automatic detection and tracking of narrow band signals,which makes sonar operators easily detect,track and classify moving targets of interest,and improves their surveillance efficiency,is a crucial function of passive sonar.However,confronted with the estimation of the number of lines and the dates of birth and death in practical application,sonar system is difficult to achieve automatic frequency line tracking efficiently.In view of complex situations,a novel method of frequency line tracking based on HMM is presented.Each frequency line is modeled as a first order Markov chain,estimated by forward and backward algorithm,and detected using sequential test.The proposed method can efficiently achieve automatic multiple frequency line tracking,and estimate dates of birth and death of target tracks.The processing results from simulation and sea trials have validated the proposed method.