As one of the basic constraints in geographical space

spatial direction relation represents geometric distribution relations of spatial objects and is described and operated by formal models in GIS.However

the existing models are incapable of distinguishing variations between two direction relations of complex objects.This paper aims to develop a computational model with high resolution in identifying the effect of parameters such as distance and shape of spatial objects on their direction relations.First

a generic quantitative approach on direction relations was proposed on the basis of the classical computation of point/point relations

in which both geometric structures and relative distribution relations of spatial objects were utilized.To obtain a reasonable qualitative description

the conversion function was further developed to link the quantitative computation with qualitative description of spatial direction relations.Numerical examples were also provided to demonstrate how to apply the proposed model into practice.Both theoretical comparison and numerical analysis indicated that the model presented in this paper is more sensitive to the parameters such as shape

size and location of spatial objects

and thus is more accurate in distinguishing different direction relations.In particular

the resulting direction relations obtained from the model satisfy three basic cognitive characteristics

including continuity

reflectivity

and asymmetry.