Abstract:

In order to solve the ineffective deficiency of the traditional trajectory distance calculation method in the indoor space, this paper puts forward a suitable method IMTSM (indoor-space moving-object trajectory similarity measurement) for indoor space trajectory similarity measurement. Firstly, according to user requirements, considering the influences of the space, time and location semantic on the trajectory similarity measurement, this paper puts forward the reconstruction algorithm, spatial distance algorithm and temporal distance algorithm of the trajectory. Then, using the relationship of positional semantics, this paper designs semantic analysis tree LSR_Tree (location semantic relation tree) which can smartly convert text similarity computation to location semantic relation computing. Besides, based on that relation, this paper proposes semantic distance algorithm which effectively reduces the deviation issue of text similarity computation. Finally, this paper transforms the trajectory distance into trajectory similarity value by means of Min-max normalization. The experimental results show effectiveness of the proposed method.

Key words: indoor space, spatio-temporal distance for trajectory, trajectory location semantic distance, trajectory similarity

摘要：

为了解决目前传统轨迹距离计算方法无法直接应用于室内空间轨迹相似性查询的问题，提出了一种适用于室内空间的轨迹相似性度量方法IMTSM（indoor-space moving-object trajectory similarity measure-ment）。首先，结合用户需求，综合考虑空间、时间、位置语义三种因素对室内移动轨迹相似性度量的影响，分别给出了轨迹重构算法、轨迹空间距离度量算法和轨迹时间距离度量算法；然后，利用位置语义之间的联系设计了位置语义分析树LSR_Tree（location semantic relation tree），将文本相似计算巧妙地转换为位置语义关系计算，并在此基础上提出了位置语义距离度量算法，有效减少了将轨迹位置语义序列作为文本序列比较的误差；最后，采用Min-max标准化处理量化轨迹距离值并转化为轨迹相似值，通过实验验证了所提方法的有效性。

关键词: 室内空间, 轨迹时空距离, 轨迹位置语义距离, 轨迹相似度