Autoencoder neural network method for curve data compression

doi:10.11947/j.AGCS.2024.20230580

Abstract

Abstract:

Vector spatial data compression stands as the most direct and effective means for reducing storage space and conserving network transmission bandwidth. This study introduces a compression and decompression model tailored for vector curve data, leveraging autoencoder neural networks. Confronting curves of varying data volume and complexities, the model initiates with segmenting, resampling processes and normalizing input vectors. Through the optimization of autoencoder structure and adjustment of the loss function, a high-precision compression-decoding module is achieved. The stability of the data is ensured through closed coordinate differences of arc segment data. Compression and restoration experiments were carried out on the 1∶1 000 000 county-level arc segments in Shanxi, Hunan, and Jiangxi provinces. The analysis reveals that the accuracy of data restoration decreases with decrease in the compression rate. When the compression rate falls below 35%, the accuracy of data restoration exhibits a fluctuating trend. Therefore, a compression rate of 25% is recommended to ensure the required level of accuracy. Comparison with Fourier series and Bézier curve fitting methods indicates advantages in compression accuracy and processing speed within specific compression rate ranges for the proposed model. Additionally, the model highlights the potential of deep learning in extracting geometric features of spatial elements. In summary, this research presents a model for vector spatial data compression based on autoencoder neural networks, demonstrating promising performance and emphasizing the potential of deep learning in spatial feature extraction.

Key words: autoencoder, vector spatial data compression, polyline simplification, map generalization

CLC Number:

P208

Pengcheng LIU, Hongran MA, Yang ZHOU, Ziqin SHAO. Autoencoder neural network method for curve data compression[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(8): 1634-1643.

Figures/Tables 9

Fig.1

Fig.2

Fig.3

Tab.1

Fig.4

Fig.5

Tab.2

Tab.3

Fig.6

References 26

[1]	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.
[2]	NAKOS B, MITROPOULOS V. Local length ratio as a measure of critical points detection for line simplification[C]//Proceedings of the 5th Workshop on Progress in Automated Map Generalization. Paris: IEEE, 2003.
[3]	SAALFELD A. Topologically consistent line simplification with the Douglas-Peucker algorithm[J]. Cartography and Geographic Information Science, 1999, 26(1): 7-18.
[4]	LI Zhilin, OPENSHAW S. Algorithms for automated line generalization on a natural principle of objective generalization[J]. International Journal of Geographical Information Systems, 1992, 6(5): 373-389.
[5]	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.
[6]	VISVALINGAM M, WHYATT J D. Line generalization by repeated elimination of points[M]//Landmarks in Mapping. Oxford: Routledge, 2017: 144-155.
[7]	WANG Zeshen, MÜLLER J C. Line generalization based on analysis of shape characteristics[J]. Cartography and Geographic Information Systems, 1998, 25(1): 3-15.
[8]	顾腾, 陈晓勇, 刘成强. 一种Douglas-Peucker与Li-Openshaw结合改进的曲线化简方法[J]. 东华理工大学学报(自然科学版), 2016, 39(4): 396-400.
	GU Teng, CHEN Xiaoyong, LIU Chengqiang. A modified line simplification method combined Douglas-Peucker with Li-Openshaw[J]. Journal of East China University of Technology (Natural Science), 2016, 39(4): 396-400.
[9]	李成名, 郭沛沛, 殷勇, 等. 一种顾及空间关系约束的线化简算法[J]. 测绘学报, 2017, 46(4): 498-506. DOI: 10.11947/j.AGCS.2017.20160546.
	LI Chengming, GUO Peipei, YIN Yong, et al. A line simplification algorithm considering spatial relations between two lines[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(4): 498-506. DOI: 10.11947/j.AGCS.2017.20160546.
[10]	LI Juncheng, ZHAO Dongbiao. An investigation on image compression using the trigonometric Bézier curve with a shape parameter[J]. Mathematical Problems in Engineering, 2013: 731648.
[11]	SARFRAZ M, ASIM M R, MASOOD A. Capturing outlines using cubic Bezier curves[C]//Proceedings of 2004 International Conference on Information and Communication Technologies: From Theory to Applications. Damascus: IEEE, 2004: 539-540.
[12]	LIU Pengcheng, LI Xingong, LIU Weibo, et al. Fourier-based multi-scale representation and progressive transmission of cartographic curves on the Internet[J]. Cartography and Geographic Information Science, 2016, 43(5): 454-468.
[13]	LIU Pengcheng, XIAO Tianyuan, XIAO Jia, et al. A multi-scale representation model of polyline based on head/tail breaks[J]. International Journal of Geographical Information Science, 2020, 34(11): 2275-2295.
[14]	吴凡. 基于小波分析的线状特征数据无级表达[J]. 武汉大学学报(信息科学版), 2004, 29(6): 488-491.
	WU Fan. Scaleless representations for polyline spatial data based on wavelet analysis[J]. Geomatics and Information Science of Wuhan University, 2004, 29(6): 488-491.
[15]	吴凡, 祝国瑞. 基于小波分析的地貌多尺度表达与自动综合[J]. 武汉大学学报(信息科学版), 2001, 26(2): 170-176.
	WU Fan, ZHU Guorui. Multi-scale representation and automatic generalization of relief based on wavelet analysis[J]. Geomatics and Information Science of Wuhan University, 2001, 26(2): 170-176.
[16]	ZHOU Qi, LI Zhilin. Use of artificial neural networks for selective omission in updating road networks[J]. The Cartographic Journal, 2014, 51(1): 38-51.
[17]	ZHOU Qi, LI Zhilin. A comparative study of various supervised learning approaches to selective omission in a road network[J]. The Cartographic Journal, 2017, 54(3): 254-264.
[18]	DU Jiawei, WU Fang, YIN Jichong, et al. Polyline simplification based on the artificial neural network with constraints of generalization knowledge[J]. Cartography and Geographic Information Science, 2022, 49(4): 313-337.
[19]	JIANG Baode, XU Shaofen, LI Zhiwei. Polyline simplification using a region proposal network integrating raster and vector features[J]. GIScience & Remote Sensing, 2023, 60(1): 2275427.
[20]	COURTIAL A, EL AYEDI A, TOUYA G, et al. Exploring the potential of deep learning segmentation for mountain roads generalisation[J]. ISPRS International Journal of Geo-Information, 2020, 9(5): 338.
[21]	DU Jiawei, WU Fang, XING Ruixing, et al. Segmentation and sampling method for complex polyline generalization based on a generative adversarial network[J]. Geocarto International, 2022, 37(14): 4158-4180.
[22]	YU Wenhao, CHEN Yujie. Data-driven polyline simplification using a stacked autoencoder-based deep neural network[J]. Transactions in GIS, 2022, 26(5): 2302-2325.
[23]	YAN Xiongfeng, YANG Min. A deep learning approach for polyline and building simplification based on graph autoencoder with flexible constraints[J]. Cartography and Geographic Information Science, 2024, 51(1): 79-96.
[24]	TONG Xiaohua, ZHOU Lejingyi, JIN Yanmin. Positional error model of line segments with modeling and measuring errors using brownian bridge[J]. Journal of Geodesy & Geoinformation Science, 2023, 6(2): 1-10.
[25]	QIAN Haizhong, ZHANG Meng, WU Fang. A new simplification approach based on the oblique-dividing-curve method for contour lines[J]. ISPRS International Journal of Geo-Information, 2016, 5(9): 153.
[26]	SANTURKAR S, TSIPRAS D, ILYAS A, et al. How does batch normalization help optimization?[J]. Advances in neural information processing systems, 2018, 31(1): 2483-2493.

省份	图形	长度范围(km)的弧段数量				汇总
省份	图形	[0，10)	[10，20)	[20，30)	[30，＋∞)	汇总
湖南省		52	150	617	23	842
江西省		39	127	475	26	667
山西省		49	124	382	35	590

压缩率/(%)	中间层神经元数量	图上坐标偏移量/mm	面积变化率/(%)	拓扑错误数量
5	10	0.492	0.822 7	25
10	20	0.243	0.274 1	20
15	30	0.174	0.224 7	15
20	40	0.127	0.172 6	13
25	50	0.112	0.157 5	5
30	60	0.119	0.170 7	4
40	80	0.102	0.149 7	5
50	100	0.130	0.169 4	2

压缩率/(%)	方法	点偏移误差/mm	处理时间/s	拓扑错误
15	本文模型	0.174	8.589	11
	傅里叶模型	0.217	15.357	202
	贝塞尔模型	0.282	20.233	19
25	本文模型	0.112	8.788	7
	傅里叶模型	0.130	25.053	150
	贝塞尔模型	0.160	18.415	12
35	本文模型	0.103	8.734	4
	傅里叶模型	0.078	36.779	96
	贝塞尔模型	0.098	15.525	16