中文摘要：

本文重点研究了空间目标间方向关系的统计建模疗法。提出的统计模型可以描述为：（1）线、面目标通过内插处理被分解成相应的基本单元；（2）空间目标间的方向关系视作这些基本单元间方向的集合；（3）利用中值（median）和分布范围（range）两种度量来完整描述方向分布的统计特性。由于此模型是从空间目标本身出发的，避免了现有模型中对空间目标的近似处地过程，从而能更精确地描述方向关系。

英文摘要：

Direction relation as a kind of spatial constraint has been recognized to be very useful for spatial query, analysis and reasoning. This paper focuses on modeling of direction relation between spatial objects. For two given spatial objects, the description result of their direction relation is to large extent dependent on the employed modeling approach. However, existing models only represent approximately the direction relation due to too coarse representation of extended spatial objects by either their minimum bounding rectangle or a point. In this paper, a statistical model is presented, in which （1） extended spatial ohjeets are decomposed into small components （i.e. suh-segments and cells）： （2） the direction relation between spatial objects is regarded as the set of the directions between these small components, i. e. a distrihution： and （3） two measures （i.e. distribution range and median direction） are utilized to describe the statistical properly of the distrihution. Indeed, the statistieal model is based upon the involved spatial objects themselves, instead of the approximation of these spatial objects in the existing models. As a result, this model is able to describe the direction relation of spatial objects more aecuratelv.