一种基于形状特征的地理实体相似性查询方法

A Method on Similarity Query of Geographic Entities Based on Shape Feature

为适应矢量空间数据库的相似性查询的应用需求,提出一种融合区域和边界的形状特征提取算法。通过地理实体的坐标,求解地理实体的质心及离散的旋转角度和质心距离序列,然后等角度间隔重采样,通过线性内插求出所得系列边界点的质心距离,建立质心距离直方图,构造以质心距离直方图、紧凑度和面积的三元组构成的形状特征描述。在此基础上,提出针对矢量空间数据的地理实体相似性查询算法。通过自主开发的GIS空间智能查询与分析平台,以全国1∶25万县市级行政区划矢量数据为数据源,实现了地理实体的相似性查询。实验结果表明,改进图像分析领域的形状分析方法并应用到矢量数据相似性查询领域是可行的,提出的方法可以满足地理实体的相似性查询应用需求。

To meet similarity query requirements for vector spatial database, this paper presents a shape feature descriptor with the fusion of regional and boundary features with its extraction algorithm. First, discrete rotation angles and a sequence of cen- troid-distances by the coordinates of the geometry are solved. Then, a centroid-distance sequence for series boundary points by e- qual interval resampling angle are determined by a linear interpolation technique. Thereupon, the centroid-distanee histogram is set. Accordingly, the descriptor with the triple of centroid-distance histogram, compactness and area is achieved. Then, an algo- rithm for geographic entities similarity query is proposed based on the descriptor, and the similarity queries for geographic enti- ties are achieved through the self-developed intelligent vector geospatial database query platform with 1 ： 250 000 nationwide county/city level administrative division vector data as experiment data set. Experimental results show that it is feasible to intro- duce shape analysis method in the filed of image analysis into the filed of geographic entities similarity query, and the proposed method can achieve geographic entity similarity query application requirements.