锥形模型的面群方向关系相似性度量方法

A similarity measurement method for direction relationship of surface groups based on cone model

针对目前群组之间的方向关系描述模型存在方向关系表达不准确和多尺度群组之间方向关系相似性度量研究的不足,该文提出一种基于锥形模型的面群方向关系表达模型,并在其基础上实现多尺度面群方向关系的相似性度量方法。首先,利用八方向锥形模型表达单个目标间的方向关系。然后,结合目标大小和目标间的距离对锥形模型方向关系计算的影响,得到两个面群之间的方向关系表达模型。最后,通过定义方向片间的“方向距离”,以交通运输最小代价为原则,计算出多尺度面群之间方向关系相似度。实验结果表明,该方法较好地克服了方向关系矩阵模型的缺陷,能较准确地计算出尺度变换后,两个面群之间的方向关系相似性。

Aiming at the inaccuracy of the representation of directional relationship among groups in the current description models and the shortcomings of the research on the similarity measurement of directional relationship among groups at multi-scale, an expression model of the direction relationship of surface groups based on conical model was proposed, and a similarity measurement method of the direction relationship of surface groups at multi-scale was implemented based on it in this paper. Firstly, the eight direction cone model was used to express the direction of the relationship between a single target. Then, considering the effect of the target size and the distance between the targets on the calculation of conical model direction relationship, the representation mode of direction relationship of surface groups was obtained. Finally, the similarity of direction relationship between surface groups at multi-scale was calculated by defining the “direction distance” of direction slice and the minimum transportation cost as the principle. Experimental results showed that this model overcomed the defects of the direction relation matrix model, which could accurately calculate the similarity of direction relationship between two surface groups after scale transformation.