A generative neural network method for road simplification

doi:10.11947/j.AGCS.2024.20230245.

Abstract

Abstract:

Road data is an important part of geospatial data due to its large quantity and high frequency of change. Road feature simplification is also one of the core technical steps of cartographic generalization and spatial data updating.Traditional methods based on data point compression, bend recognition and existing machine learning algorithms have problems of poor stability, weak controllability and low degree of automation in road simplification. Based on the theory of combining visual thinking and syntactic pattern. This paper uses the feature mining ability of deep learning algorithm to introduce generative artificial neural network model into the field of road simplification.Firstly, the road data to be simplified is transformed into sequence data, and the sequence features are extracted to construct the feature data set. Secondly, the Seq2Seq coding model constructed by the gated recurrent unit (GRU) neural network is used to embed large-scale road data to form high-dimensional semantic coding. The simplified small-scale road data is generated by decoding the semantic coding. Finally, the effectiveness and applicability of the model were evaluated according to four indexes: compression rate of arc segment, change rate of length, bend of curve and buffer limit difference.The experimental results show that the proposed model can be applied to road shape simplification, enrich road simplification methods, and promote the comprehensive and intelligent development of map drawing.

Key words: generative neural network, road simplification, deep learning, syntactic pattern recognition natural language processing, Seq2Seq coding model, cartographic generalization

CLC Number:

P283

Piao LUO, Junkui XU, Fang WU, Yakun LÜ, Qingwen ZHUANG. A generative neural network method for road simplification[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(10): 2007-2020.

Figures/Tables 20

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Fig.7

Tab.1

Fig.8

Fig.9

Fig.10

Tab.2

Fig.11

Fig.12

Tab.3

Fig.13

Tab.4

Tab.5

Fig.14

Tab.6

References 33

[1]	武芳, 巩现勇, 杜佳威. 地图制图综合回顾与前望[J]. 测绘学报, 2017, 46(10):1645-1664. DO1:. doi: 10.11947/j.AGCS.2017.20170287
	WU Fang, GONG Xianyong, DU Jiawei. Overview of the research progress in automated[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10):1645-1664. DO1:. doi: 10.11947/j.AGCS.2017.20170287
[2]	DOUGLAS D H, PEUCKER T K. Algorithms for the reduction of the number of points required to represent a digitized line or its caricature[J]. Cartographica: the International Journal for Geographic Information and Geovisualization, 1973, 10(2):112-122.
[3]	LI Zhilin, OPENSHAW S. Algorithms for automated line generalization 1 on a natural principle of objective generalization[J]. International Journal of Geographical Information Systems, 1992, 6(5):373-389.
[4]	TEH C H, CHIN R T. On the detection of dominant points on digital curves[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1989, 11(8):859-872.
[5]	邓敏, 陈杰, 李志林, 等. 曲线简化中节点重要性度量方法比较及垂比弦法的改进[J]. 地理与地理信息科学, 2009, 25(1):40-43.
	DENG Min, CHEN Jie, LI Zhilin, et al. An improved local measure method for the importance of vertices in curve simplification[J]. Geography and Geo-Information Science, 2009, 25(1):40-43.
[6]	朱鲲鹏, 武芳, 王辉连, 等. Li-Openshaw算法的改进与评价[J]. 测绘学报, 2007, 36(4):450-456.
	ZHU Kunpeng, WU Fang, WANG Huilian, et al. Improvement and assessment of Li-Openshaw algorithm[J]. Acta Geodaetica et Cartographica Sinica, 2007, 36(4):450-456.
[7]	王荣, 闫浩文, 禄小敏. Douglas-Peucker算法全自动化的多尺度空间相似关系方法[J]. 地球信息科学学报, 2021, 23(10):1767-1777.
	WANG Rong, YAN Haowen, LU Xiaomin. Automation of the Douglas-Peucker algorithm based on spatial similarity relations[J]. Journal of Geo-Information Science, 2021, 23(10):1767-1777.
[8]	WANG Zeshen, MÜLLER J C. Line generalization based on analysis of shape characteristics[J]. Cartography and Geographic Information Systems, 1998, 25(1):3-15.
[9]	毋河海. 数字曲线拐点的自动确定[J]. 武汉大学学报(信息科学版), 2003, 28(3):330-335.
	WU Hehai. Automatic determination of inflection point and its applications[J]. Geomatics and Information Science of Wuhan University, 2003, 28(3):330-335.
[10]	郭庆胜, 黄远林, 章莉萍. 曲线的弯曲识别方法研究[J]. 武汉大学学报(信息科学版), 2008, 33(6):596-599.
	GUO Qingsheng, HUANG Yuanlin, ZHANG Liping. The method of curve bend recognition[J]. Geomatics and Information Science of Wuhan University, 2008, 33(6):596-599.
[11]	艾廷华, 郭仁忠, 刘耀林. 曲线弯曲深度层次结构的二叉树表达[J]. 测绘学报, 2001, 30(4):343-348.
	AI Tinghua, GUO Renzhong, LIU Yaolin. A binary tree representation of curve hierarchical structure in depth[J]. Acta Geodaetica et Cartographic Sinica, 2001, 30(4):343-348.
[12]	翟仁健, 武芳, 朱丽, 等. 曲线形态的结构化表达[J]. 测绘学报, 2009, 38(2):175-182.
	ZHAI Renjian, WU Fang, ZHU Li, et al. Structured representation of curve shape[J]. Acta Geodaetica et Cartographica Sinica, 2009, 38(2):175-182.
[13]	杜佳威, 武芳, 李靖涵, 等. 采用多元弯曲组划分的线要素化简方法[J]. 计算机辅助设计与图形学学报, 2017, 29(12):2189-2196.
	DU Jiawei, WU Fang, LI Jinghan, et al. Line simplification method based on multi-bends groups division[J]. Journal of Computer-Aided Design & Computer Graphics, 2017, 29(12):2189-2196.
[14]	钱海忠, 何海威, 王骁, 等. 采用三元弯曲组划分的线要素化简方法[J]. 武汉大学学报(信息科学版), 2017, 42(8):1096-1103.
	QIAN Haizhong, HE Haiwei, WANG Xiao, et al. Line feature simplification method based on bend group division[J]. Geomatics and Information Science of Wuhan University, 2017, 42(8):1096-1103.
[15]	AI Tinghua, KE Shu, YANG Min, et al. Envelope generation and simplification of polylines using Delaunay triangulation[J]. International Journal of Geographical Information Science, 2017, 31(2):297-319.
[16]	郑春燕, 郭庆胜, 胡华科. 基于蚁群优化算法的线状目标简化模型[J]. 测绘学报, 2011, 40(5):635-638.
	ZHENG Chunyan, GUO Qingsheng, HU Huake. The simplification model of linear objects based on ant colony optimization algorithm[J]. Acta Geodaetica et Cartographica Sinica, 2011, 40(5):635-638.
[17]	段佩祥, 钱海忠, 何海威, 等. 基于支持向量机的线化简方法[J]. 武汉大学学报(信息科学版), 2020, 45(5):744-752, 783.
	DUAN Peixiang, QIAN Haizhong, HE Haiwei, et al. A line simplification method based on support vector machine[J]. Geomatics and Information Science of Wuhan University, 2020, 45(5):744-752, 783.
[18]	杜佳威, 武芳, 朱丽, 等. 图形、图像融合利用的集成学习智能化简方法及其在岛屿岸线化简中的应用[J]. 测绘学报, 2022, 51(3):373-387. DOI:. doi: 10.11947/j.AGCS.2022.20210135
	DU Jiawei, WU Fang, ZHU Li, et al. An ensemble learning simplification approach based on multiple machine-learning algorithms with the fusion using of raster and vector data and a use case of coastline simplification[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(3):373-387. DOI:. doi: 10.11947/j.AGCS.2022.20210135
[19]	DU Jiawei, WU Fang, XING Ruixing, et al. An automated approach to coastline simplification for maritime structures with collapse operation[J]. Marine Geodesy, 2021, 44(3):157-195.
[20]	YAN Xiongfeng, AI Tinghua, YANG Min, et al. Graph convolutional autoencoder model for the shape coding and cognition of buildings in maps[J]. International Journal of Geographical Information Science, 2021, 35(3):490-512.
[21]	武芳, 杜佳威, 钱海忠, 等. 地图综合智能化研究的发展与思考[J]. 武汉大学学报(信息科学版), 2022, 47(10):1675-1687.
	WU Fang, DU Jiawei, QIAN Haizhong, et al. Overview of research progress and reflections in intelligent map generalization[J]. Geomatics and Information Science of Wuhan University, 2022, 47(10):1675-1687.
[22]	朱强, 武芳, 钱海忠, 等. 一种顾及认知规律的曲线弯曲识别方法[J]. 辽宁工程技术大学学报(自然科学版), 2014, 33(4):521-526.
	ZHU Qiang, WU Fang, QIAN Haizhong, et al. An identification method of line curves based on cognitive laws[J]. Journal of Liaoning Technical University (Natural Science), 2014, 33(4):521-526.
[23]	戴汝为. 模式识别的一类属性文法[J]. 自动化学报, 1983, 9(2):90-98.
	DAI Ruwei. Pattern recognition is a kind of attribute grammar[J]. Journal of Automation, 1983, 9(2):90-98.
[24]	SUTSKEVER I, VINYALS O, LE Q V. Sequence to sequence learning with neural networks[J]. Advances in Neural Information Processing Systems, 2014, 4(1):3104-3112.
[25]	CHO K, VAN MERRIENBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[C]//Proceedings of 2014 Conference on Empirical Methods in Natural Language Processing. Doha: Association for Computational Linguistics, 2014.
[26]	ZHAO Rui, WANG Dongzhe, YAN Ruqiang, et al. Machine health monitoring using local feature-based gated recurrent unit networks[J]. IEEE Transactions on Industrial Electronics, 2018, 65(2):1539-1548.
[27]	LUONG M T, PHAM H, MANNING C D. Effective approaches to attention-based neural machine translation[J]. Computer Science, 2015. DOI:. doi: 10.18653/v1/D15-1166
[28]	BAHDANAU D, CHO K, BENGIO Y. Neural machine translation by jointly learning to align and translate[J]. Computer Science, 2014. DOI:. doi: 10.48550/arXiv.1409.0473
[29]	KÖRDING K P, WOLPERT D M. The loss function of sensorimotor learning[J]. Proceedings of the National Academy of Sciences of the United States of America. 2004, 101(26):9839-9842.
[30]	KINGMA D, BA J. Adam: a method for stochastic optimization[J]. Computer Science, 2014. DOI:. doi: 10.48550/arXiv.1412.6980
[31]	KESKAR N S, MUDIGERE D, NOCEDAL J, et al. On large-batch training for deep learning: generalization gap and sharp minima[J]. Computer Science, 2016. DOI:. doi: 10.48550/arXiv.1609.04836
[32]	何海威, 钱海忠, 段佩祥, 等. 线要素化简及参数自动设置的案例推理方法[J]. 武汉大学学报(信息科学版), 2020, 45(3):344-352.
	HE Haiwei, QIAN Haizhong, DUAN Peixiang, et al. Case based reasoning method for line element simplification and automatic parameter setting[J]. Journal of Wuhan University (Information Science Edition), 2020, 45(3):344-352.
[33]	武芳, 朱鲲鹏. 线要素化简算法几何精度评估[J]. 武汉大学学报(信息科学版), 2008, 33(6):600-603.
	WU Fang, ZHU Kunpeng. Geometric accuracy assessment of linear features' simplification algorithms[J]. Geomatics and Information Science of Wuhan University, 2008, 33(6):600-603.

学习率	MSE	训练耗时
0.000 5	0.007 82	33 h 10 min
自适应调整	0.004 06	33 h 21 min
0.000 015 625	0.035 1	33 h 38 min

网络层数	MSE	训练耗时	最优Acc迭代	最优Acc耗时
1	0.004 06	33 h 21 min	280	31 h 06 min
2	0.004 49	42 h 39 min	292	41 h 27 min
3	0.005 68	52 h 10 min	300	52 h 10 min

神经元数	MSE	训练耗时	Acc/(%)	最优Acc迭代	最优Acc耗时
64	0.005 61	30 h 00 min	90.41	300	30 h 00 min
128	0.004 06	33 h 21 min	94.82	280	31 h 06 min
256	0.004 68	40 h 58 min	95.76	215	29 h 16 min

化简方法	删除弧段数	压缩率/(%)
D-P	1920	89.42
L-O	1154	53.75
Seq2Seq网络	1061	49.42
参考样例	1051	48.95

化简方法	长度变化比	曲折度	缓冲区误差
D-P	0.998 4	0.518 9	0.981 8
L-O	0.996 1	0.671 0	0.995 5
Seq2Seq网络	0.998 4	0.763 8	0.989 0