Journal of Geo-information Science >
Pattern Recognition of Regular Buildings with Unbalanced Sample Size
Received date: 2021-06-16
Revised date: 2021-07-16
Online published: 2022-03-25
Supported by
National Natural Science Foundation of China(41761088)
National Natural Science Foundation of China(42061060)
Excellent platform support of Lanzhou Jiaotong University(201806)
Tianyou innovation team of Lanzhou Jiaotong University(TY202001)
Copyright
It is of profound significance in the fields of multi-scale map expression and digital mapping to recognize the architectural pattern intelligently by integrating spatial relations, geometry, semantics, and other features. Applying graph convolution neural network to intelligent recognition of building pattern can overcome the shortcomings of traditional methods, such as relying on artificial experience to set parameters and making rules too strict. However, this method often has the problem of sample proportion imbalance, which easily leads to the category with a small number of samples can not be properly identified. In this paper, the building centriod is used as the sample to cluster to obtain the building groups. Secondly, the building centriod in the same group is used as the node to construct the graph structure of the building group of Delaunay triangulation. The feature selection of the graph node can describe the area, size, direction, and other related indicators of the geometric characteristics of the building group. Thirdly, the building groups graph structure with small number of samples is enhanced by oversampling the graph structure. Then the data before and after the enhancement of the architectural group graph structure with small number of samples are input into the graph convolution neural network model for training, and the performance of the model is evaluated with several evaluation indexes such as ROC curve. The experimental results show that the recognition accuracy of the model for buildings with lesser number of samples is significantly improved after the structure of buildings with small number of samples is enhanced.
LING Zhenfei , LIU Tao , DU Ping , ZHANG Yaorong , YANG Guolin , SUO Xuhong . Pattern Recognition of Regular Buildings with Unbalanced Sample Size[J]. Journal of Geo-information Science, 2022 , 24(1) : 63 -73 . DOI: 10.12082/dqxxkx.2022.210340
表1 建筑物指标选取Tab. 1 index selection of building |
变量 | 指标 | 计算方法 | 公式编号 | 描述 |
---|---|---|---|---|
位置 | 质心 | (2) | 所有顶点的算术平均位置,其中N是顶点的数量 | |
尺寸 | 周长 | Perimeter=Pb | (3) | 建筑物周长 |
面积 | Area=Ab | (4) | 建筑物面积 | |
均半径 | Mean radius= | (5) | 建筑各顶点到质心的平均距离[12] | |
方向 | SBRO WSWO | 最小边界矩形方向具有小公差的建筑物的墙壁方向[13] | ||
紧凑度 | (6) | 面积和周长的平方关系[14] | ||
分形度 | (7) | 面积与周长的对数关系[14] | ||
形状 | 延伸率 | (8) | SBR建筑的长宽比[15] | |
凹度 | (9) | 建筑与等面积圆的交合处面积比[16] | ||
密度 | 面积比率 | (10) | 建筑面积与其影响面积的比率(IA) |
表2 规则建筑群增强前数据分布Tab. 2 Data distribution before enhancement of the regular building group pattern (个) |
类别 | 训练集数量 | 验证集数量 | 测试集数量 | 总量 |
---|---|---|---|---|
不规则样本 | 633 | 211 | 211 | 1055 |
规则样本 | 62 | 21 | 21 | 104 |
总量 | 695 | 232 | 232 | 1159 |
表3 规则建筑群增强后数据分布Tab. 3 Data distribution after enhancement of the regular building group pattern (个) |
类别 | 训练集数量 | 验证集数量 | 测试集数量 | 总量 |
---|---|---|---|---|
不规则样本 | 635 | 212 | 212 | 1059 |
规则样本 | 616 | 205 | 206 | 1027 |
总量 | 1251 | 417 | 418 | 2086 |
表4 模型准确率对比Tab. 4 Comparison of model accuracy (%) |
方法 | 训练集准确率 | 测试集准确率 |
---|---|---|
增强前GCNN | 91.00 | 91.12 |
增强后GCNN SVM | 97.2090.76 | 93.7890.08 |
表5 规则建筑群识别准确率Tab. 5 Recognition accuracy of regular buildings (%) |
方法 | 训练集准确率(规则建筑群识别) | 测试集准确率(规则建筑群识别) |
---|---|---|
增强前GCNN | 0.80 | 0.60 |
增强后GCNN | 72.82 | 73.54 |
[1] |
徐永洋. OpenStreetMap城市建筑物数据质量评价方法研究[D]. 武汉:中国地质大学, 2019.
[
|
[2] |
艾廷华, 郭仁忠. 基于格式塔识别原则挖掘空间分布模式[J]. 测绘学报, 2007, 36(3):302-308.
[
|
[3] |
巩现勇. 顾及分布特征和道路网约束的居民地综合方法研究[D]. 郑州:战略支援部队信息工程大学, 2017.
[
|
[4] |
蔡娇楠. 顾及格式塔原则的街区建筑物MST聚类[D]. 西安:长安大学, 2018.
[
|
[5] |
|
[6] |
|
[7] |
晏雄锋. 深度卷积学习支持下的建筑物模式分析[D]. 武汉:武汉大学, 2019.
[
|
[8] |
巩现勇, 武芳, 钱海忠, 等. 建筑群多连通直线模式的参数识别方法[J]. 武汉大学学报·信息科学版, 2014, 39(3):335-339.
[
|
[9] |
刘鹏程, 林文辉, 许小峰. 一种傅立叶描述子支持下的建筑群模式识别方法[J]. 华中师范大学学报(自然科学版), 2018, 52(5):750-756.
[
|
[10] |
巩现勇, 武芳. 基于图匹配的城市建筑群典型字母型分布的识别[J]. 武汉大学学报·信息科学版, 2018, 43(1):159-166.
[
|
[11] |
孙前虎. 基于多约束的建筑群聚类方法研究[D]. 长沙:中南大学, 2011.
[
|
[12] |
|
[13] |
|
[14] |
|
[15] |
王米琪, 艾廷华, 晏雄锋, 等. 图卷积网络模型识别道路正交网格模式[J]. 武汉大学学报·信息科学版, 2020, 45(12):1960-1969.
[
|
[16] |
杨永春, 张从果, 刘治国. 快速集聚发展过程中的河谷型城市的空间整合与规划——以兰州市为例[J]. 干旱区地理, 2004, 27(4):603-609.
[
|
[17] |
罗蒙, 黄晓东, 江波, 等. Logistic回归联合ROC曲线模型预测新型冠状病毒肺炎患者发生危重症的风险[J]. 中草药, 2020, 51(20):5287-5292.
[
|
/
〈 | 〉 |