基于方向关系矩阵的空间方向相似性定量计算方法

摘要

This article introduces a new model for direction relations between spatial objects at multiple scales and a corresponding similarity assessment method .The model is an improvement of direction relation matrix ,which quantitatively models direction relations on object scale ,and by means of the solution of the Transportation Problem to solve the minimum conversion cost between direction matrices ,namely distance between a pair of matrices ,thus quantified the difference between a pair of directions ,finally obtain the similarity values between arbitrary pairs of spatial objects and compare the results .Experiments on calcu‐lating similarity between objects at different scales show that the presented method is efficient ,accurate , and capable of obtaining results consistent with human cognition .%介绍了一种多尺度空间对象的方向关系表达模型以及基于该模型的方向相似度度量方法。该方向关系模型对方向关系矩阵模型进行了改进，根据空间对象的形状定量描述空间对象之间的方向关系；借鉴平衡传输问题的解决方法计算方向矩阵间最小转换代价，即方向矩阵间的距离，从而量化方向对间的差异，最终获得任意尺度空间对象的方向相似度并对其进行比较。对不同尺度空间对象的方向相似性的试验表明，该方法简单可行且不失精度，结果符合人类认知。