中文摘要：

为解决现有空间对象形状相似性匹配准确率较低的问题,提出一种应用三角形划分的形状相似性匹配方法。该方法按形状主方向对面状空间对象进行分割,按串联、并联和组合形式对空间对象进行三角形划分,准确描述面状空间对象的形状特征,度量空间对象间的形状相似性。通过形状数据集匹配、不同年份面状水系图层匹配和矢量地图草图检索,测试本方法的形状检索性能,并和其他空间对象形状匹配方法进行对比。实验结果表明,本方法具有更高的形状检索准确率。三角形划分形状匹配方法具有平移、旋转、尺度不变性和较强的形状描述识别能力。

英文摘要：

Existing spatial object shape similarity matching methods are not accurates. To solve this problem, a spatial object shape matching method based on triangular division is proposed. This method segments the areal spatial object through the main direction of the object shape and divides the object shape into triangles into a series form, parallel connection form, and combined form. This method describes the shape features of the areal spatial object exactly and measures the shape similarity of spatial objects. The matching on the shape data set, the matching on areal water in different years and sketch retrieval on a vector map are used to test the retrieval performance of this method. This method is compared to other spatial object shape matching methods. Experimental results show that this method has higher retrieval accuracy. The spatial object shape matching method based on triangular division is invariant to translation, rotation, scaling, and has a strong capacirty to describe and recognize shapes.