摘要：

针对行为轨迹还原过程中观察序列状态缺失、无法对终端轨迹进行精确还原的问题，提出一种基于隐马尔可夫模型的行为轨迹还原算法。利用基站布局的空间相关性，在不考虑缺失观察状态的情况下，对隐马尔可夫模型求解过程中的局部概率进行修订，还原出轨迹序列。性能分析和仿真结果表明，状态倾向度越大，轨迹还原成功率越高，当状态倾向度取0.8时，轨迹还原成功率在90%左右。

关键词: 行为轨迹, 状态倾向度, 轨迹还原, 状态缺失, 局部概率, 隐马尔可夫模型

Abstract:

This paper proposes a behavior trajectory restoration algorithm for observation sequence state missing problem, which leeds to terminal trajectory restoration inaccurately. The algorithm utilizes base station layout’s spatial correlation and revises the partial probability of the solution process of the Hidden Markov Models(HMM) to restore the track sequence without considering the missing observation states. Performance analysis and simulation results show that the greater the degree of state propensity is, the higher the success rate of trajectory restoration is. When the degree of state propensity is 0.8, the success rate of trajectory restoration is about 90 percent.

Key words: behavior trajectory, state propensity degree, trajectory restoration, state missing, partial probability, Hidden Markov Models(HMM)

中图分类号: 

• TN92