简单曲线无量纲形状相似度计算方法
闫浩文(1969—),男,甘肃民勤人,博士,教授,博士生导师,研究方向为地图自动综合、空间分析等。E-mail: yanhw@mail.lzjtu.cn |
收稿日期: 2023-07-03
修回日期: 2023-07-25
网络出版日期: 2023-12-05
基金资助
国家自然科学基金项目(41930101)
Calculation of Nondimensional Shape Similarity between Simple Curves
Received date: 2023-07-03
Revised date: 2023-07-25
Online published: 2023-12-05
Supported by
National Natural Science Foundation of China(41930101)
曲线图形形状相似度的计算是地图学、图形学、几何学中的基础性理论问题之一,虽然目前有机器学习的方法可以计算曲线形状相似度,但这种方法往往依赖于大样本曲线对,其效率并不高。为了解决这一问题,本文提出了一种直接计算曲线形状相似度的方法。原理如下:① 对2条曲线进行位移、旋转和比例变换,找出能够使两条曲线叠置在一起后重合程度最大或平均距离最小的位置; ② 根据曲线的顶点和交集,将两条曲线划分为不同的子区域。然后,根据Gestalt心理学的邻近性原理,在每个子区域内计算图形的形状相似度; ③ 将各个子区域内图形的形状相似度进行加权求和,得到2个简单曲线目标的形状相似度。通过心理学实验研究,验证了本文提出的方法计算得到的形状相似度与人们的空间认知相符,具有一定的适用场景。该方法不仅可以直接计算曲线形状的相似度,而且避免了对大样本曲线对的依赖,提高了计算效率。因此,该方法在地图学、图形学和几何学等领域具有一定的应用前景。
闫浩文 , 杨维芳 , 禄小敏 , 诸天舒 , 马犇 , 殷硕硕 . 简单曲线无量纲形状相似度计算方法[J]. 地球信息科学学报, 2023 , 25(12) : 2418 -2426 . DOI: 10.12082/dqxxkx.2023.230368
Calculation of shape similarity between curves is one of the most fundamental and theoretical problems in cartography, graphics, and geometry. Although existing machine learning methods can be used to calculate curve shape similarity, they often rely on extensive sets of sample curves, leading to a low efficiency. To address this issue, this paper proposes a method for directly calculating shape similarity between simple curves. First, two curves are moved, rotated, and scaled to obtain the optimal position where the mean distance between the two curves is the least. Second, the two curves are divided into a number of subsections based on their intersections of the curves. Third, the shape similarity within each subsection (i.e., two sub-curves) is calculated by the principle of proximity in Gestalt. Finally, the shape similarity of the two curves can be obtained by calculating the weighted shape similarity of all subsections. The proposed method is validated through the psychological experiments, and the results show that the calculated shape similarity aligns with human spatial cognition, indicating its practical applicability in specific scenarios. Moreover, the proposed method not only directly calculates curve shape similarity but also eliminates the reliance on a large number of curve samples, resulting in increased computational efficiency. The method presented in this paper provides a more efficient and direct tool for calculating curve shape similarity and holds promise for applications in various fields such as cartography, graphics, and geometry.
Key words: curve; shape similarity; rotation; scaling; psychological experiments; calculation; geometry
图6 形状相似度是否合理的心理测试Fig. 6 A psychological test of whether the shape similarity is plausible |
表1 形状相似度是否合理的心理测试结果Tab. 1 The results of the psychological experiments on whether the shape similarity is plausible (%) |
图形组别 | 形状相似度计算结果认同度 | ||
---|---|---|---|
专业人员 | 非专业人员 | 全部人员 | |
1 | 99.48 | 100.00 | 99.66 |
2 | 99.48 | 98.99 | 99.32 |
3 | 93.81 | 96.97 | 94.88 |
4 | 97.94 | 96.97 | 97.61 |
5 | 94.85 | 95.96 | 95.22 |
6 | 95.36 | 97.98 | 96.25 |
7 | 96.39 | 98.99 | 97.27 |
8 | 98.97 | 98.99 | 98.98 |
9 | 99.48 | 98.99 | 99.32 |
10 | 100.00 | 97.98 | 99.32 |
平均 | 97.58 | 98.18 | 97.78 |
图7 形状相似度变化是否合理的心理测试Fig. 7 A psychological test of whether the changes in shape similarity is plausible |
表2 形状相似度准确与否的心理实验结果Tab. 2 The results of the psychological experiments on whether the shape similarity is accurate (%) |
图形组别 | 形状相似度计算结果认同度 | ||
---|---|---|---|
专业人员 | 非专业人员 | 全部人员 | |
1 | 90.21 | 92.93 | 91.13 |
2 | 91.24 | 95.96 | 92.83 |
3 | 91.24 | 92.93 | 91.81 |
4 | 89.18 | 95.96 | 91.47 |
平均 | 90.47 | 94.45 | 91.81 |
[1] |
郭仁忠. 空间分析[M]. 武汉: 武汉测绘科技大学出版社,1997.
[
|
[2] |
陆利正, 何歆, 凌海雅, 等. 空间曲线的特征识别与高质量非均匀采样[J]. 计算机辅助设计与图形学学报, 2022, 34(1):18-24.
[
|
[3] |
魏明山, 朱明明, 刘光花, 等. 一种特征点和曲线相似度的目标识别方法[J]. 雷达科学与技术, 2022, 20(6):688-696,704.
[
|
[4] |
穆国旺, 张志伟, 臧婷, 等. 基于和参考曲线相似性的B样条曲线延拓[J]. 计算机辅助设计与图形学学报, 2018, 30(9):1705-1711.
[
|
[5] |
|
[6] |
安晓亚, 刘平芝, 金澄, 等. 手绘地图开域空间方向关系检索法[J]. 测绘学报, 2017, 46(11):1899-1909.
[
|
[7] |
仇阿根. 基于分布式内存计算的空间数据近似查询处理方法[J]. 测绘学报, 2017, 46(12):2044.
[
|
[8] |
行瑞星, 武芳, 巩现勇, 等. 建筑群组合直线模式识别的模板匹配方法[J]. 测绘学报, 2021, 50(6):800-811.
[
|
[9] |
李晨, 申德荣, 朱命冬, 等. 一种对时空信息的kNN查询处理方法[J]. 软件学报, 2016, 27(9):2278-2289.
[
|
[10] |
艾廷华, 帅赟, 李精忠. 基于形状相似性识别的空间查询[J]. 测绘学报, 2009, 38(4):356-362.
[
|
[11] |
闫浩文. 空间相似关系[M]. 北京: 科学出版社, 2022.
[
|
[12] |
|
[13] |
孙琛琛, 申德荣, 寇月, 等. 面向实体识别的聚类算法[J]. 软件学报, 2016, 27(9):2303-2319.
[
|
[14] |
李平, 阳武, 吴佳英, 等. 基于相似度预测的WSN数据收集算法[J]. 软件学报, 2014, 25(s1):93-102.
[
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
程绵绵, 孙群, 李少梅, 等. 多尺度点群广义Hausdorff距离及在相似性度量中的应用[J]. 武汉大学学报(信息科学版), 2019, 44(6):885-891.
[
|
[23] |
张建业, 潘泉, 张鹏, 等. 基于斜率表示的时间序列相似性度量方法[J]. 模式识别与人工智能, 2007, 20(2):271-274.
[
|
[24] |
张萍, 李必军, 郑玲, 等. 一种基于改进LCSS的相似轨迹提取方法[J]. 武汉大学学报·信息科学版, 2020, 45(4):550-556.
[
|
[25] |
|
[26] |
程绵绵, 孙群, 徐立, 等. 面轮廓线相似性和复杂性度量及在化简中的应用[J]. 测绘学报, 2019, 48(4):489-501.
[
|
[27] |
|
[28] |
|
/
〈 |
|
〉 |