Research on multi-scale vector road matching model based on ISOD descriptor

doi:10.21203/rs.3.rs-4816732/v1

In the data processing of geographic information, the matching of road data at different scales is crucial. Due to scale differences, road features can change, posing a challenge to multi-scale matching,.Spatial relationships are essential for multi-scale matching because they remain stable at different scales. In this paper, we propose an improved the summation product of orientation and distance (ISOD) descriptor, which combines features such as pinch chain code and curvature variance with similarity metrics, such as length, direction, and Hausdorff distance, to construct an integrated similarity metric model for multi-scale road matching. The experiments proved that the model achieved 94.75% and 93.34% check accuracy and completeness in road data matching at scales of 1:50,000 and 1:10,000. The model also achieved 86.39% and 94.06% check accuracy and completeness in road data matching at scales of 1:250,000 and 1:50,000, respectively. This proves the effectiveness and practicality of the method. The ISOD descriptor and the integrated similarity metric model in this paper provide an effective method for multi-scale road data matching, which aids the integration and fusion of geographic information data and is significantly valuable when applied in the fields of intelligent transport and urban planning.

spatial relationships

ISOD descriptors

pinch chain codes

curvature variance

multi-scale road networks

Currently, vector spatial data matching methods mainly use geometric matching, topological matching and semantic matching as the basis of entity discrimination¹. For road network vector data matching, line feature matching is its main aspect. Line features are usually matched using geometric features such as distance, shape and angle.

In the field of road matching, multi-scale road network matching methods usually focus on geometric similarity while ignoring semantic similarity, and the weights and thresholds often rely on subjective settings, which may lead to mis-matching. In order to improve the matching efficiency, Bingjiao Wu^2–3, proposed a fusion of geometric and semantic similarity matching methods and created an innovative set of subjective–objective combinations of semantic matching algorithms for road networks, thus reducing human intervention and improve the accuracy of matching. However, the existing road network matching method still requires human control and cannot achieve full automation, and it is susceptible to the interference of irrelevant roads when processing scale difference data, which leads to mis-matching. To solve this problem, Qin Yuluo et al.⁴ designed a hierarchical progressive accurate matching strategy, based on preliminary matching, which meticulously takes into account geometric and topological features. In addition, Luo⁵ proposed a global matching optimisation method for buildings based on road network constraints, while Guo Xuan et al.⁶ improved matching accuracy by constructing an expense network flow model, evaluating street similarity in terms of three dimensions—spatial distance, directionality and geometric morphology—and introducing a global optimal matching. On the other hand, Wang et al.⁷ proposed an incremental path-based matching method for complex urban road network data to improve accuracy and speed. In addition, Li⁸ developed an MM algorithm based on the semantic fusion of in-vehicle images, which is particularly suitable for parallel road scenarios and can be applied in areas such as unmanned navigation and map updating.

In the field of algorithm improvement for multi-source data matching, Li Hao⁹ proposed a multi-model fusion strategy. He constructed and trained multiple machine learning sub-models to judge data matching scenarios and used the dynamic classifier selection technique to flexibly select the best sub-models according to the geometric features and similarity of roads, which subsequently improved the accuracy of the algorithm. Binbin Xie¹⁰ designed a new road matching algorithm with the help of the Hidden Markov Model, which effectively solved the problem of mis-matching caused by the large deviation of isolated GPS tracking points. In addition, Wu¹¹ introduced a new Voronoi-diagram-based multi-scale road network matching formula (VAMRN) and confirmed that it outperforms the two existing methods in terms of generality and matching quality. Singh et al.¹²classified and analysed the performance of map matching algorithms and found that, for online map matching, the algorithm based on the Hidden Markov Model exhibits higher accuracy in comparison with other methods.

(3) In road network data matching, researchers have made significant progress with the help of system or network models. Haocheol Jiang¹³ proposed a road matching model based on graph neural networks that adaptively learns and applies underlying laws in spatial data without the need to construct complex similarity metrics. Xinglin Xu¹⁴, on the other hand, developed a set of techniques to enhance the speed of data docking and refreshing in order to optimise the efficiency of matching mobile vehicle paths with road networks. Zhao Ye¹⁵ designed a novel end-to-end road beacon image alignment and pairing technique to achieve direct mapping from input to output through deep learning techniques, which can be used to solve many of the problems encountered when using traditional methods. In addition, Yan et al.¹⁶ proposed a framework for analysing the matching of traffic flow and road resources using large-scale mobile phone data. Moreover, Cheng et al.¹⁷ designed a road location transformation method and provided a deep-learning-based semantic matching model for roads. Wang et al.¹⁸ introduced a multi-scale dynamic alignment technique for detecting and updating different matching information between data sources. Shen et al.¹⁹ proposed a method based on YOLOv3 (You Only Look Once version 3) to enhance cross-scale detection capability and the focus on valuable regions. Their method addresses the urgent need for multi-scale detection and its components play an important role in road target detection. Xie²⁰ designed a real-time artificial intelligence road detection system based on binocular vision sensors, aiming to improve the reliability of road condition detection. The system can run on a low-power edge computing platform and transmit processing results to the cloud via Internet of Things (IoT) devices. In terms of road information extraction from remotely sensed images, Gong²¹ proposed an algorithm based on semantic segmentation, which can still achieve high-precision matching under greater interference. Meanwhile, Lei²² et al. introduced a method based on GIS to generate a high-resolution map model, which can be used for lane-level map matching and shows good performance. Finally, Ma et al.²³ proposed a new algorithm that uses a small number of GPS measurements to determine the initial probability of the hidden state, thus improving the accuracy of pedestrian map matching.

The types of road differences at different scales for the same geographic entity can be classified as geometric differences, topological differences, semantic differences, and other differences. Among these, other differences include differences in coordinate systems, differences in data models, and differences in data formats. The attributes of data at different scales usually lack uniquely identifiable information. Therefore, geometric matching is generally used to identify homonymous entities^24–26. Geometric differences in road datasets at different scales can be categorised as spatial-location-based, orientation-based, shape-based, and length-based^27–29.

In this study, three road network datasets with different scales of 1:10,000, 1:50,000, and 1:250,000 from different sources and phases in Xiangshan County, Zhejiang Province, are used as the experimental data, among which 160 articles in the dataset used a scale of 1:250,000, 2,376 used a scale of 1:50,000, and 6359 used a scale of 1:250,000.As shown in Fig. 1.

Overview of Experimental Models

This paper presents seven road matching models applied to match an experimental road dataset. Among these models, length, orientation, and Hausdorff distance serve as common fundamental matching metrics shared across all. Additionally, the ISOD descriptor, angle chain code, and curvature variance^30–32are integrated into the road matching models individually or in combination to further enhance their matching capabilities.Table.1 introduces the composition of each model.

Table 1

The index composition of multiple models
Model names	Model composition
Models I	Composed of basic indicators fused with ISOD descriptors.
Models Cr	Composed of basic indicators fused with angle chain codes.
Models Cv	Composed of basic indicators fused with curvature variance.
Models ICr	Composed of basic indicators fused with both ISOD descriptors and angle chain codes.
Models ICv	Composed of basic indicators fused with both ISOD descriptors and curvature variance.
Models CrCv	Composed of basic indicators fused with both angle chain codes and curvature variance.
Models ICrCv	Composed of basic indicators fused with ISOD descriptors, angle chain codes, and curvature variance.

Road segmentation and landmark extraction

In this paper, experiments were conducted on two sets of road datasets: 1:250,000 and 1:50,000 scale datasets and 1:50,000 and 1:100,000 scale datasets. The reference datasets were at scales of 1:250,000 and 1:50,000, respectively. When extracting landmarks in the dataset with a scale of 1:250,000, intersections with a number of nodes greater than 2 are directly selected as landmarks due to the small scale and the small number of road entries. The larger the scale, the more detailed the roads in the dataset. In the dataset with a scale of 1:50,000, this paper extracts road intersections with a node number greater than 3 as landmarks. The changes in the number of roads before and after road segmentation in different datasets are shown in Table 2.

Table 2

Changes in the number of roads before and after road segmentation in different datasets
Count Scale	1:250000	1:50000	1:10000
Pre-split	160	2376	6359
Post-split	376	5358	14410

Table 3 shows the number of landmark points in the 1:250,000 and 1:50,000 road datasets. Fig. 2 and Fig. 3 respectively show the changes in the number of intersections before and after extracting landmarks from the 1:250000 and 1:50000 road datasets.

Table. 3 Changes in the number of intersections before and after extracting landmarks from road datasets

Data set scale	Pre-screening landmarks	After filtering the landmarks
1：250000	332	169
1：50000	4078	458

Model indicator weights and matching results

Considering that the span of the 1:10,000 and 1:250,000 scales is large and the accuracy of the experimental results is low, in this paper, the road datasets at scales of 1:50,000 and 1:250,000, as well as those at scales of 1:10,000 and 1:50,000 are named as Group A and Group B, respectively, for road matching experiments. The road dataset with a scale of 1:250,000 is the reference road dataset in Group A, and the road dataset with a scale of 1:50,000 is the reference road dataset in Group B. In this paper, a total of seven road matching models are selected to be applied in the matching of the experimental road dataset. Among them, length, direction and Hausdorff distance are used as the common matching indexes for these seven road datasets, while ISOD descriptor, angle chain code, and curvature variance are added to the road matching models either individually or in combination. In this paper, the weight of each metric in each model is determined by controlling a single variable. When determining the weights for a particular model, the weights of one of the metrics are changed first, the weights of the other metrics are set and kept constant, and then multiple comparison experiments are conducted. The weight corresponding to the indicator with the best matching result is the weight of that indicator in that model. Subsequently, the weight of the next indicator is changed, the weights of the other indicators remain unchanged, and then several comparison experiments are conducted to select the weight with the best matching result. By comparison, the weights of each indicator in the model can be determined for matching in groups A and B, respectively. The weights of each indicator in the seven models were normalised. Tables 4 and 5 show the distribution of each metric for the seven models in the Group A and Group B experiments, respectively. The matching results are shown in Fig. 4 and Fig. 5.

Table 4

Distribution of Index Weights for 7 Models in Group A Experiment
Matching Models	Similarity indicators
Matching Models	Length	Directions	Hausdorff distance	ISOD	Corner chain code	variance of curvature
Models I	0.2	0.2	0.85	0.8
Models Cr	0.2	0.2	0.95		0.7
Models Cv	0.2	0.3	0.8			0.3
Models ICr	0.1	0.2	0.9	0.8	0.7
Models ICv	0.1	0.1	0.9	0.9		0.1
Models CrCv	0.2	0.2	0.9		0.6	0.1
Models ICrCv	0.2	0.3	0.9	0.7	0.6	0.1

Table 5

Distribution of Index Weights for 7 Models in Group B Experiment
Matching Models	Similarity indicators
Matching Models	Length	Directions	Hausdorff distance	ISOD	Corner chain code	variance of curvature
Models I	0.3	0.25	0.9	0.9
Models Cr	0.3	0.3	0.75		0.65
Models Cv	0.35	0.4	0.7			0.2
Models ICr	0.2	0.3	0.85	0.8	0.75
Models ICv	0.3	0.3	0.8	0.75		0.2
Models CrCv	0.25	0.25	0.8		0.5	0.2
Models ICrCv	0.3	0.35	0.8	0.9	0.7	0.15

Firstly, this paper compares and analyses the matching results of models I, Cr, and Cv. The similarity metrics shared by these three models are length, direction, and Hausdorff distance, and the differences are that the three models combine ISOD descriptors, pinch chain codes, and curvature variance, respectively. The output graph and data of the comprehensive measurement model combining four similarity measures are shown in Table 6, Table 7, and Fig. 6, and are compared and analyzed by comparing the distribution and evaluation metrics of mismatched roads and target roads.

Table 6

Evaluation results of Models I, Cr, and Cv in Experiments A
Matching Models	Precision(%)	Recall(%)	F-score值(%)
Models I	86.39	94.06	90.06
Models Cr	85.15	89.55	89.29
Models Cv	76.60	91.25	83.29

Table 7

Evaluation results of Models I, Cr, and Cv in ExperimentsB
Matching Models	Precision(%)	Recall(%)	F-score值(%)
Models I	94.75	93.34	94.04
Models Cr	95.20	92.98	94.08
Models Cv	86.48	86.46	86.47

In the matching experiments in Group A, all three models have higher check-full rates than check-accurate rates. With the three common indicators unchanged, model I combined with ISOD has the highest F-score value, i.e., the best matching effect. The scale of the reference roads in Group A is 1:250,000, and due to the small scale, most of the roads are main roads, and a very small portion of them are branch roads. As shown in the derived graphs, the mis-matching cases are mainly distributed in main roads, and the mis-matches are mainly distributed in branch roads. In the matching experiment of group B, the detection rate of all three models is higher than the detection rate. In the case that the three indicators remain unchanged, the model Cr combined with the angle chain code has the highest F-score value, i.e., the matching effect is the best. The scale of the reference road in Group B is 1:50,000 (a larger scale), the road display is more detailed, and the zoning of the trunk road and the reference road is significant. The mis-matching is mainly distributed in the branch road, and incorrect matching is mainly distributed in the trunk road. When matching between small-scale road datasets, we can consider combining spatial relationship indicators based on geometric similarity indicators, i.e., ISOD descriptors, to match road datasets, creating a better matching effect. When matching between large-scale road datasets, the geometric similarity index can be considered to be combined with the angle chain code, and the matching effect is better.

Subsequently, this paper outputs the output plots of the matching results of the integrated metric models—ICr, ICv and CrCv—which combine the five similarity metrics and data (shown in Table 8, Table 9 and Fig. 7), performing comparative analyses by comparing the distributions of the mis-matched roads among the target roads, as well as evaluating the metrics.

Table 8

Evaluation results models ICr, ICv and CrCv in Experiments A
Matching Models	Precision(%)	Recall(%)	F-score value(%)
Models ICr	76.81	87.35	81.74
Models ICv	81.96	93.06	87.16
Models CrCv	82.73	89.48	85.97

Table 9

Evaluation results models ICr, ICv and CrCv in gExperiments B
Matching Models	Precision(%)	Recall(%)	F-score value(%)
Models ICr	94.91	93.06	93.98
Models ICv	88.86	88.76	88.81
Models CrCv	90.81	90.01	90.41

In the road dataset matching experiments in Group A, the models—ICr, ICv, and CrCv—all have higher check-perfect rates than check-accurate rates. The model ICv combining ISOD descriptors and curvature variance has the highest F-score value, i.e., the best matching effect, when the three metrics in common are unchanged. Mis-matches are mainly distributed on trunk roads, while mis-matches are more evenly distributed on trunk and branch roads, but mis-matches are more concentrated in areas with dense branch roads. In the road dataset matching experiments in group B, the check accuracy rates of models—ICr, ICv, and CrCv—are all higher than the check accuracy rate and the model ICr, which combines the ISOD descriptor. The angle chain code has the highest F-score value, i.e., it has the best matching effect, while the three shared metrics remain unchanged. Mis-matches are distributed in both main roads and branch roads, while incorrect matches are mainly distributed in branch roads.

When performing the matching of small-scale road datasets while keeping length, distance, and direction constant, the combination of ISOD descriptor and curvature variance can be considered for matching since it has the best matching effect. When matching large-scale road datasets, the ISOD descriptor and the angle chain code has the best matching effect. Spatial relationships can used for the matching of both large-scale or small-scale road datasets since the matching effect is more accurate.

Table 10

Evaluation results of Model ICrCv in Experiments A
Matching Models	Precision(%)	Recall(%)	F-score value(%)
Models ICrCv	83.57	90.17	86.74

Table. 11 Evaluation results of Model ICrCv in Experiments B

Matching Models	Precision(%)	Recall(%)	F-score value(%)
Models ICrCv	91.01	90.54	90.27

Subsequently, the matching results of the model ICrCv in Groups A and B experiments are compared and analysed in this paper (as shown in Table 10, Table 11 and Fig. 8). With the increase of similarity index, the matching model is more constrained for road matching, and ISOD descriptors, pinch chain codes and curvature variance are added at the same time while keeping the distance, length and direction unchanged. By comparing the matching results, the model is more suitable for road matching between large-scale road datasets. In Group A experiments, mis-matching cases are distributed in the main roads and incorrect matching cases are mainly distributed in the branch roads. In Group B experiments, the mis-matches are mainly distributed in branch roads. Incorrect matches are distributed in both main roads and branch roads, and these matches are more concentrated in areas with dense branch roads.

Ultimately, this study compares the matching results of the seven models in the two groups of road datasets and analyses the matching results in terms of the precision, recall, and F-score values. The overall matching results for the road datasets in Groups A and B are shown in Tables 12 and 13.

Table 12

Matching results of Group A road dataset
Matching Models	FN	FP	TP	Precision(%)	Recall(%)	F-score(%)
Models I	69	172	1092	86.39	94.06	90.06
Models Cr	105	157	900	85.15	89.55	89.29
Models Cv	102	325	1064	76.60	91.25	83.29
Models ICr	129	269	891	76.81	87.35	81.74
Models ICv	83	245	1113	81.96	93.06	87.16
Models CrCv	107	190	910	82.73	89.48	85.97
Models ICrCv	102	184	936	83.57	90.17	86.74

Table 13

Matching results of Group B road dataset
Matching Models	FN	FP	TP	Precision(%)	Recall(%)	F-score(%)
Models I	582	452	8158	94.75	93.34	94.04
Models Cr	602	402	7971	95.20	92.98	94.08
Models Cv	1148	1146	7333	86.48	86.46	86.47
Models ICr	598	430	8021	94.91	93.06	93.98
Models ICv	965	955	7620	88.86	88.76	88.81
Models CrCv	843	769	7598	90.81	90.01	90.41
Models ICrCv	804	760	7695	91.01	90.54	90.27

As shown in Fig. 9, Group A matching experiments were conducted on road datasets with scales of 1:250,000 and 1:50,000, where Model I, Model Cr and Model ICrCv had higher checking accuracy rates of 86.39%, 85.15% and 83.57%, respectively, with Model 1 having the highest checking accuracy. Model I, Model Cv and Model ICv had higher check accuracy rates of 94.06%, 91.25% and 93.06%, respectively. Among these, Model I had the highest check accuracy rate. In Group A experiments, all seven groups of experimental results had lower check accuracy than check completeness. As a result, Model I has the best matching effect when performing matching between small-scale road datasets. In other words, the comprehensive similarity model consisting of four similarity indicators—distance, direction, length and ISOD descriptor—was applied to match roads between small scales, and its check-perfect rate was higher than the check-accurate rate, i.e., the cases of mis-matches were lower than the cases of incorrect matches.

This study also evaluates and analyses the matching results of Group B road datasets. As shown in Fig. 10, in performing the matching of Group B with scales of 1:50,000 and 1:10,000, Model I, Model Cr and Model ICr had higher checking accuracy rates of 94.75%, 95.20% and 94.91%, respectively, with Model Cr having the highest checking accuracy. Model I, Model Cr and Model ICr had higher detection rates of 93.34%, 92.98% and 93.06%, respectively, with Model I having the highest detection rate. Since the highest checking accuracy and the highest checking completeness are not from the same matching model, they need to be combined with the F-score value for further quality evaluation. In the Group B experiment, all seven groups of matches passed through the check accuracy rate higher than the check completeness rate.

As shown in Fig. 11, among the seven sets of experiments in Group A, Model I had the highest F-score value of 90.06%, and Model Cr had the next highest F-score value of 89.29%. Among the seven sets of experiments in Group B, Model Cr had the highest F-score value of 94.08%, and Model I had the second highest F-score value of 94.04%. In summary, when matching between road datasets, a comprehensive similarity metric model that combines length, Hausdorff distance, direction and ISOD descriptors can be used for matching, and its comprehensive matching effect is the best. Overall, the matching effect of all seven matching models in large-scale road datasets is better than that of small-scale road datasets. Therefore, the larger the scale of the road dataset, the better the effect of matching using the comprehensive similarity model.

With the rapid development of society and the economy, various fields are implementing increasingly higher requirements for the quality and availability of road data. As an important part of geospatial information, the accuracy and real-time performance of road data are crucial for urban planning, traffic management, disaster emergency response, and other fields. In this experiment, based on three geometric similarity indicators—length, Hausdorff distance and direction—ISOD descriptor, angle chain code, and curvature variance are combined to construct a comprehensive similarity metric model for multi-scale road matching experiments at scales of 1:10,000 and 1:50,000, as well as 1:50,000 and 1:250,000. By comparing the results of the experiments, it is found that the matching effect on the large-scale road network is better than that of the small-scale road network matching as a whole, which is because the large-scale road network has more detailed and precise road information and provides more road matching points. In the same set of experiments, with the change in the number of model indicators, the overall matching effect will also change since the model needs to consider changes in information complexity. The experiments found that the comprehensive similarity metric model constructed through multiple indicators can more accurately reflect the actual characteristics of the road to improve matching accuracy. Due to the complexity and diversity of road data sources, the use of integrated evaluation models reduces the reliance on a single indicator, improves robustness, and is more adaptable to different environments.

This experiment is based on three geometric metrics, and innovatively integrates three metrics—ISOD descriptor, angle chain code and curvature variance—to construct a fusion model with several different metrics, enriching the comprehensive similarity metric modelling approach in the field of multi-scale road matching. However, there are also shortcomings in this experiment. Although two new metrics are introduced, the complexity of the matching process increases along with the increase in the number of model metrics, which reduces the efficiency of matching. In addition to this, the selection of metrics for evaluating the model is a challenge. How to effectively fuse the information of different metrics and the different matching methods may have a significant impact on the results.

Data preparation

Firstly, we acquire the required geospatial vector data and road vector data at various scales for the experiment. Next, we extract road data at different scales from the vector data using the 'Select By Location' functionality in ArcGIS. Following that, we proceed with road segmentation. We open the editor, utilize the 'Split Lines At Intersections' tool to segment intersecting roads, and then convert the road segments into points using the 'Feature to Point' tool, resulting in multiple road marking points. Due to the different scales, the number of road intersections varies. In the 1:50,000 scale map, we select intersections with more than 3 nodes as landmarks, whereas in the 1:250,000 scale map, intersections with more than 2 nodes are chosen as landmarks. Finally, we create a network dataset, link it with the landmark data, and generate a file named Join_Output. By opening the attribute table of this file, we can select corresponding landmarks based on different scales for further analysis or visualization.

Model construction and analysis

We built models based on distance, orientation, and Hausdorff distance with the addition of ISOD descriptors, angular chain codes, and curvature variance. These features were combined to form three independent models (I, Cr, Cv) and three composite models (ICr, ICv, CrCv). Finally, we fused all six metrics into the ICrCv model.After comparing these models with a reference group on the dataset, we analysed the correct and incorrect rates of matches using accuracy, recall and F1 scores.

Data availability

The data that support the findings of this study are available from the corresponding author upon request.

Acknowledgements

This work was supported in part by the Geological Survey Project of China Geological Survey (No. DD20220956), in part by the Major Project of High-Resolution Earth Observation System of China (No. GFZX0404130304), and in part by the Shandong Province Culture and Tourism Research Project (No. 23WL(Y)53), and also in part by the Zibo City Social Science Planning Research Project (No. 2023ZBSK041).

Author information

Authors and Affiliations

School of Civil Engineering and Geomatics, Shandong University of Technology, Zibo, Shandong province, 255049, China.

Yuefeng Lu,Ying Sun,Yu Yan

The State Key Laboratory of Efficient Utilization of Arid and Semi-arid Arable Land in Northern China (the Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing, 100081, China).

Miao Lu

Author contributions statement

Y.L. and Y.S. came up with this idea, with Y.L. and M.L. guiding the experiment throughout. Y.S. and Y.Y. conducted the experiments and completed the initial draft of the manuscript. Y.L. and M.L. then guided the revision of this initial draft. Finally, all authors participated in the writing and finalization of the manuscript.

Corresponding authors

Correspondence to Yu Yan,Miao Lu.

Competing Interests

The authors declare no competing interests.

Zhang, Z.C. Spatial Data Orgnization of Multi-scale Linear Objects Based on Features. Master's Thesis, Wuhan University, Wuhan, China (2005).
Wu, B.J. An approach for Multi-scale Road Networks Data Matching Method Based on Geometric and Semantic Similarity. Master's Thesis, Lanzhou Jiao tong University, Lanzhou, China (2022).
Wu, B.J., Wang Z.H. & Yang, F. A semantic similarity computational model for multi-scale road network matching. Surv. Mapp. Sci.47,166-173 (2022).
Qin, Y.L., Song, W.D., Zhang, Y.Z. & Sun, X.L. Matching method for road network considering geometric features and topological continuity. Bull. Sur. Mapp.55-60 (2021).
Luo, G.W., Qin, K. X. Global optimisation matching method for multi-representation buildings constrained by road network. J. Comput. Methods Sci. Eng.23, 2413-2424 (2023).
Guo, X., Zhang, Y.F. & Qiu, Z.H. Multi-index road network matching method based on minimum cost network flow. Eng. Surv. Mapp. 32, 13-19 (2023).
Wang, H.Y., Liu, Y.G., Li, S.H., Liang, Bo.& He, Z.Y. A Path Increment Map Matching Method for High-Frequency Trajectory," IEEE Trans. Intell. Transp. Syst.24, 10948-10962 (2023).
Li, B.Z., Wang, M.Q, Cai, Z.L, Su, S.L. & Kang, M. J.VIS-MM: a novel map-matching algorithm with semantic fusion from vehicle-borne images. Eng. Surv. Mapp. 37, 1069-1098 (2023).
Li, H. Research on Constructing Road Network Hierarchy Model and Its Application on Road Matching. M.S. Thesis, Information Engineering University , Zhengzhou, China (2020).
Xie, B.B. Research on Road Matching Algorithm of Taxi OD Time Series Trajectory Based on Hidden Markov Model. M.S. Thesis, Chang'an University, Xi'an, China (2020).
Wu, J. et al. A new Voronoi diagram-based approach for matching multi-scale road networks," J. Geogr. Syst. 25, 265-289 (2023).
Singh, S., Singh, J., Goyal, S., El Barachi, M. & Kumar, M.Analytical review of map matching algorithms: analyzing the performance and efficiency using road dataset of the indian subcontinent. J. Geogr. Syst. 30, 4897-4916 (2023).
Jiang, H.Z. Research On Adjacent scale Road Data Matching Based On Graph Neural Network. M.S. Thesis, Henan University, Kaifeng, China (2022).
Xu, X.L. Fast Map Matching Between Floating Car Data and Road Network and Its Application in Road Update. M.S. Thesis, Wuhan University, Wuhan, China (2021).
Zhao, Y. Cross-Domain Aerial-Road Registration And Matching Based On Deep Learning. M.S. Thesis, Southeast University, Nanjing, China (2020).
Yan, X. et al. Revealing spatiotemporal matching patterns between traffic flux and road resources using big geodata-A case study of Beijing. Cities.127, 103754 (2022).
Cheng, R. & Chen, J.A location conversion method for roads through deep learning-based semantic matching and simplified qualitative direction knowledge representation. Eng. Appl. Artif. Intell. 104, 104400 (2021).
Wang, Y.H., Yu, B.B., Zhu, F.X., Zhang, J.C. & Huang, C. Research on intelligent design method of ship multi-deck compartment layout based on improved taboo search genetic algorithm. Ocean Eng. 25, 3155-3173 (2021).
Shen, L. et al. Improved YOLOv3 model with feature map cropping for multi-scale road object detection. Meas. Sci. Technol. 34, 045406 (2023).
Xie, Q., Hu, X., Ren, L., Qi, L. & Sun, Z. A binocular vision application in IoT: Realtime trustworthy road condition detection system in passable area. Eng. Appl. Artif. Intell 19, 973-983(2022).
Gong, Z. Road Vector Data Extraction and Matching Positioning Application Based on Remote Sensing Image. M.S. thesis, Huazhong University of Science and Technology, Wuhan, China (2020).
Lei, T., Xiao, G.Y. & Yin, X.H. Targeting Lane-Level Map Matching for Smart Vehicles: Construction of High-Definition Road Maps Based on GIS. Appl. Sci. 13, 862 (2023).
Ma, S. & Lee, H.A Practical HMM-Based Map-Matching Method for Pedestrian Navigation. Int. Conf. Inf. Networking. 806-811 (IEEE).
Zhao, B.A study on multi-scale vector map objects matching method and its application. PhD dissertation, Central South University, Changsha, China (2011).
Xu, F., Deng, M., Zhao, B. & Chen, J.A detailed investigation on the methods of object matching. J. Geo-Inf.Sci. 11, 657-663 (2009).
Song, W., Keller, J. M., Haithcoat, T. L. & Davis, C. H.Relaxation‐based point feature matching for vector map conflation.Tran.GIS. 15, 43-60 (2011).
Wang, Y., Chen, D., Zhao, Z., Ren, F. & Du, Q.A back‐propagation neural network‐based approach for multi‐represented feature matching in update propagation.Tran.GIS. 19, 964-993 (2015).
Fan, H., Yang, B., Zipf, A. & Rousell.A polygon-based approach for matching OpenStreetMap road networks with regional transit authority data. Int. J. Geogr. Inf. Sci.30, 748-764 (2016).
Han, J., Kamber, M. & Mining, D.Concepts and techniques. Morgan kaufmann. 340, 94104-93205 (2006).
Wen,Q. Research on Multi-scale Road Network Vector Data Matching Method. M.S. thesis, Shandong University of Technology, Zibo, China (2021).
Zhao, Y., Chen, Y.Q. Included Angle Chain:A Method for Curve Representation. J. Software. 300-307 (2004).
Haynes, R. et al. District variations in road curvature in England and Wales and their association with road-traffic crashes. Environ Plann A. 39, 1222-1237 (2007).

No competing interests reported.

Supplementarymaterials.docx

Research on multi-scale vector road matching model based on ISOD descriptor

Status:

Version 1

Abstract

Figures

Introduction

Overview of Experimental Data

Overview of Experimental Models

Results

Road segmentation and landmark extraction

Model indicator weights and matching results

Discussion

Methods

Data preparation

Model construction and analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1