用双切比雪夫方法近似含噪音移动对象轨迹

Approximating Inaccurate Moving Object Trajectories with Bi-chebyshev Method

在移动对象数据库中需要存储大量移动对象的历史轨迹。为了降低存储开销,同时提高轨迹查询的效率,研究者们提出了很多基于时间序列的方法对轨迹序列进行压缩近似及索引。但是这些方法不能用于不精确的轨迹数据。本文针对含噪音的轨迹数据提出了一种新的近似算法。该方法充分利用了轨迹位置数据和速度数据的导数关系,在不增加计算复杂度的情况下,能够更好地处理不精确的轨迹。在相同的压缩比下,用双切比雪夫方法重建的轨迹比现有方法更加接近移动对象的真实轨迹。

Approximating and compressing inaccurate moving object trajectories is a key topic in moving object database research. This paper proposes a new method, called Bi-Chebyshev, to solve the problem. Different from existing methods, the Bi-Chebyshev method makes use of velocity data as subsidiary information to improve the accuracy of trajectory approximation. The method is based on a quadratic optimization model, to which a numerical solution is provided. In experimental comparison, the Bi-Chebyshev method demonstrates its superiority in both reconstruction accuracy and robustness against noise. The feature coefficients generated by Bi-Chebyshev transformation can be used for indexing purpose in the same way as in other methods. The complexity of computing Bi-Chebyshev coefficients is quadratic on the number of coefficients and linear on the length trajectory data, comparable to other global methods.