Road Network Information Research Articles

SDUNet: Road extraction via spatial enhanced and densely connected UNet

Extracting road maps from high-resolution optical remote sensing images has received much attention recently, especially with the rapid development of deep learning methods. However, most of these CNN based approaches simply focused on multi-scale encoder architectures or multiple branches in neural networks, and ignored some inherent characteristics of the road surface. In this paper, we design a novel network for road extraction based on spatial enhanced and densely connected UNet, called SDUNet. SDUNet aggregates both the multi-level features and global prior information of road networks by combining the strengths of spatial CNN-based segmentation and densely connected blocks. To enhance the feature learning about prior information of road surface, a structure preserving model is designed to explore the continuous clues in the spatial level. Experimental results on two benchmark datasets show that the proposed method achieves the state-of-the-art performance, compared with previous approaches for road extraction. Code will be made available on https://github.com/MrStrangerYang/SDUNet.

High-resolution inventory of atmospheric emissions from transport, industrial, energy, mining and residential activities in Chile

Abstract. This study presents the first high-resolution national inventory of anthropogenic emissions for Chile (Inventario Nacional de Emisiones Antropogénicas, INEMA). Emissions for the vehicular, industrial, energy, mining and residential sectors are estimated for the period 2015–2017 and spatially distributed onto a high-resolution grid (approximately 1 km×1 km). The pollutants included are CO2, NOx, SO2, CO, VOCs (volatile organic compounds), NH3 and particulate matter (PM10 and PM2.5) for all sectors. CH4 and black carbon are included for transport and residential sources, while arsenic, benzene, mercury, lead, toluene, and polychlorinated dibenzo-p-dioxins and furan (PCDD/F) are estimated for energy, mining and industrial sources. New activity data and emissions factors are compiled to estimate emissions, which are subsequently spatially distributed using census data and Chile's road network information. The estimated annual average total national emissions of PM10 and PM2.5 during the study period are 191 and 173kt a−1 (kilotons per year), respectively. The residential sector is responsible for over 90 % of these emissions. This sector also emits 81 % and 87 % of total CO and VOC, respectively. On the other hand, the energy and industry sectors contribute significantly to NH3, SO2 and CO2 emissions, while the transport sector dominates NOx and CO2 emissions, and the mining sector dominates SO2 emissions. In general, emissions of anthropogenic air pollutants and CO2 in northern Chile are dominated by mining activities as well as thermoelectric power plants, while in central Chile the dominant sources are transport and residential emissions. The latter also mostly dominates emissions in southern Chile, which has a much colder climate. Preliminary analysis revealed the dominant role of the emission factors in the final emission uncertainty. Nevertheless, uncertainty in activity data also contributes as suggested by the difference in CO2 emissions between INEMA and EDGAR (Emission Database for Global Atmospheric Research). A comparison between these two inventories also revealed considerable differences for all pollutants in terms of magnitude and sectoral contribution, especially for the residential sector. EDGAR presents larger emissions for most of the pollutants except for CH4 and PM2.5. The differences between both inventories can partly be explained by the use of different emission factors, in particular for the residential sector, where emission factors incorporate information on firewood and local operation conditions. Although both inventories use similar emission factors, differences in CO2 emissions between both inventories indicate biases in the quantification of the activity. This inventory (available at https://doi.org/10.5281/zenodo.4784286, Alamos et al., 2021) will support the design of policies that seek to mitigate climate change and improve air quality by providing policymakers, stakeholders and scientists with qualified scientific spatially explicit emission information.

Road-aware Indexing for Trajectory Range Queries

Answering spatio-temporal range queries (RQs) on trajectory databases, i.e., finding all trajectories that intersect given ranges, is crucial in many real-world applications. Various kinds of indexes have been proposed to accelerate RQs. However, existing indexes typically use Euclidean distance to prune irrelevant regions without concerning the underlying road network information. Nevertheless, as vehicle trajectories are generated on road network edges, the road network could be seen as meta knowledge of trajectories and be used to index and query trajectories. To this end, we propose RP-Tree, a <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</u> oad network-aware <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</u> artition <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">tree</u> to support efficient RQs. The basic idea is partitioning a road network graph into hierarchical subgraphs and generate a balanced tree structure, where each tree node maintains its associated trajectories. We compactly index the spatio-temporal information of trajectories on the corresponding road network edges. Then, we design efficient search algorithms to support RQs by pruning irrelevant trajectories through subgraph range borders associated with RP-Tree nodes. Last but not least, we scale RP-Tree to very large datasets by devising approximate algorithms with bounded confidence at an interactive speed. Experimental results on three real-world datasets from Porto, Chengdu, and Beijing show that our method outperform baselines by 1 to 2 orders of magnitude.

A Lane Level Bi-Directional Hybrid Path Planning Method Based on High Definition Map

A global reference path generated by a path search algorithm based on a road-level driving map cannot be directly used to complete the efficient autonomous path-following motion of autonomous vehicles due to the large computational load and insufficient path accuracy. To solve this problem, this paper proposes a lane-level bidirectional hybrid path planning method based on a high-definition map (HD map), which effectively completes the high-precision reference path planning task. First, the global driving environment information is extracted from the HD map, and the lane-level driving map is constructed. Real value mapping from the road network map to the driving cost is realized based on the road network information, road markings, and driving behavior data. Then, a hybrid path search method is carried out for the search space in a bidirectional search mode, where the stopping conditions of the search method are determined by the relaxation region in the two search processes. As the search process continues, the dimension of the relaxation region is updated to dynamically adjust the search scope to maintain the desired search efficiency and search effect. After the completion of the bidirectional search, the search results are evaluated and optimized to obtain the reference path with the optimal traffic cost. Finally, in an HD map based on a real scene, the path search performance of the proposed algorithm is compared with that of the simple bidirectional Dijkstra algorithm and the bidirectional BFS search algorithm. The results show that the proposed path search algorithm not only has a good optimization effect, but also has a high path search efficiency.

Double Covering 모형을 활용한 인천광역시 소방시설 최적위치 선정에 관한 연구

The purpose of this study is to find the optimal locations of fire fighting facilities in Incheon to prevent the spread of accidental fires and efficiently extinguish them. In the event of a large fire difficult to cope with in one fire fighting facility, additional support from nearby facilities will be necessary. The model proposed in this study firstly ensures that all demand nodes in a network will be able to be reached from the minimum number of facilities in a given time. At the same time, we locate the given number of facilities so that they can cover as many demand nodes as possible multiple times. This problem can be considered as a facility location problem which aims at maximizing the number of demand nodes that are covered at least twice while minimizing the number of facilities needed. The proposed model is applied to conduct a case study using road network information and administrative district data in Incheon.

A novel method for the detection of road intersections and traffic rules using big floating car data

The information of road intersection and traffic rules plays an essential role in transportation system especially for people travelling to large, unfamiliar cites. Moreover, cities are growing fast and it is difficult to keep up with the changing frequency of road network information although road map is periodically updated to provide reliable information. In order to tackle this problem, an automatic method to detect road intersection and traffic rules using a floating car trajectory dataset is proposed. The turning trajectories using a newly proposed algorithm employing flexible spatial, temporal, and logical constraint rules were first extracted. Then a Two-trajectory Intersect Angle-based (TIA) method was proposed to locate the candidate points at road intersection. Finally the intersections (junctions) were detected using a cluster-based technique. Furthermore, turning rules at road intersections were extracted automatically by clustering the angles both before and after making a turn. The effectiveness of our method was demonstrated using big floating car trajectory dataset collected in Fuzhou, China. Additional comparisons and analysis were also conducted to confirm identification results. Our findings showed that a total 333 at-grade intersections were detected automatically with more than 94% in accuracy, and turning rules identification reached 85% in accuracy by manually comparing the results.

A multi-constraint combined method for road extraction from airborne laser scanning data

Airborne laser scanning (ALS) is a state-of-the-art technique for fast and accurate acquisition of road network information. In this paper, an ALS data-based road extraction method, with four well-designed steps namely pre-processing, intensity-based filtering, quadrant-based region growing, and road candidate regions extraction is proposed. The contributions of this paper are mainly concentrated in last three steps, where radiometric, geometric and statistical constraints are combined to differentiate road and non-road points. The proposed method is straightforward to implement, where complex cases, such as removal of attached areas to the road, extraction of road surfaces surround the irregular data gaps left by vehicles and trees are effectively dealt. The method performance was evaluated using two datasets having complex cases and road network was extracted at average completeness, correctness and quality of 84.3%, 93% and 79.2%, respectively. The method achieves significant improvement in comparison with several state-of-the-art methods.

Map matching for travel route identification based on Earth Mover's Distance algorithm using wireless cell trajectory data

Precise travel route identification is the guiding basis of feasible urban traffic planning and modeling. A range of map matching (MM) algorithms have been studied in previous research to integrate coordinate-based GPS data with digital road network information to identify personal travel route. However, these methods significantly depend on the high-density positioning data, which are usually unavailable or costly for transportation planning projects. Fortunately, mobile phones under 3 G, 4 G and 5 G networks can generate massive cell-based travel trajectory data with almost no additional cost. This data is potentially valuable for travel route identification if proper MM algorithms can be developed. This paper proposes an MM method to detect travel route by using handoff trajectory data from the cellular phone network. First, a wireless communication simulation model for dynamic cellular handoff-based traffic monitoring is developed for handoff data collection. Second, the Earth Mover’s Distance (EMD) algorithm is used to identify travel route by finding the time-space relationship among the cellular handoff patterns of the road network. Finally, by comparing the performance of the proposed method with classical sequence similarity algorithms, results indicate that the EMD-based MM algorithm is much more efficient for travel route identification, and it can detect small spacing parallel roads with a high accuracy.

An enhanced global context-aware network for road extraction from high-resolution imagery

ABSTRACT Currently, the automation of road segmentation in high-resolution satellite imagery is inadequate, and a part of the road network information is affected by the loss of spatial features and lack of global context information. To overcome this, we propose a deep convolutional neural network called the enhanced global context-aware network (EGC-Net). The EGC-Net employs a newly enhanced global context-aware module, which adds convolution kernels of different scales based on atrous spatial pyramid pooling (ASPP). In addition, the feature aggregation and self-attention modules further improve the accuracy of road segmentation. Our experimental results show that the performance of EGC-Net is superior to that of state-of-the-art methods. When tested using the Massachusetts Roads Dataset, the EGC-Net achieved an F1-score (evaluation metric for the harmonic mean of the precision and recall metrics) of 0.8234, accuracy of 0.9466, and the intersection of the prediction and ground truth regions over their union (IoU) of 0.8195. Overall, EGC-Net performs better than other deep learning algorithms under different evaluation indexes. The ablation study indicates that the IoU score of our enhanced global context-aware module is higher than that of ASPP by 0.0207, and each component is required to obtain the best road detection results.

TaxiInt: Predicting the Taxi Flow at Urban Traffic Hotspots Using Graph Convolutional Networks and the Trajectory Data

Taxi flow is an important part of the urban intelligent transportation system. The accurate prediction of taxi flow provides an attractive way to find the potential traffic hotspots in the city, which helps to avoid serious traffic congestions by taking effective measures in advance. The current prediction of taxi flow and its impact on urban transportation are closely related to the passenger origin-destination (OD) information. However, high-quality OD information is not always available. To address this problem, a prediction model, named as TaxiInt, is proposed in this study. Different from other density-clustering-based approaches, neural network, or OD information based models, TaxiInt predicted the taxi flow using the trajectory data of taxis. The spatial features and temporal features of each road were extracted using a graph convolutional network, which was trained with the road network information and the trajectory data. The experiments carried on a real taxi dataset showed the validity of our model. It can predict the taxi flow at a given urban intersection with high accuracy.

Research on Vehicle-Road Co-Location Method Oriented to Network Slicing Service and Traffic Video

The development of 5G network slicing technology, combined with the application scenarios of vehicle–road collaborative positioning, provides end-to-end, large-bandwidth, low-latency, and highly reliable flexible customized services for Internet of Vehicle (IoV) services in different business scenarios. Starting from the needs of the network in the business scenario oriented to co-location, we researched the application of 5G network slicing technology in the vehicle–road cooperative localization system. We considered scheduling 5G slice resources. Creating slices to ensure the safety of the system, provided an optimized solution for the application of the vehicle–road coordinated positioning system. On this basis, this paper proposes a vehicle–road coordinated combined positioning method based on Beidou. On the basis of Beidou positioning and track estimation, using the advantages of the volumetric Kalman model, a combined positioning algorithm based on CKF was established. In order to further improve the positioning accuracy, vehicle characteristics could be extracted based on the traffic monitoring video stream to optimize the service-oriented positioning system. Considering that the vehicles in the urban traffic system can theoretically only travel on the road, the plan can be further optimized based on the road network information. It was preliminarily verified by simulation that this research idea has improved the relative single positioning method.

Road Extraction from High Resolution Remote Sensing Images Based on Vector Field Learning.

Accurate and up-to-date road network information is very important for the Geographic Information System (GIS) database, traffic management and planning, automatic vehicle navigation, emergency response and urban pollution sources investigation. In this paper, we use vector field learning to extract roads from high resolution remote sensing imaging. This method is usually used for skeleton extraction in nature image, but seldom used in road extraction. In order to improve the accuracy of road extraction, three vector fields are constructed and combined respectively with the normal road mask learning by a two-task network. The results show that all the vector fields are able to significantly improve the accuracy of road extraction, no matter the field is constructed in the road area or completely outside the road. The highest F1 score is 0.7618, increased by 0.053 compared with using only mask learning.

RTVEMVS: Real-time modeling and visualization system for vehicle emissions on an urban road network

To improve the real-time and precision of vehicle emission monitoring and estimation for supporting variable and speedy response on vehicle emission control work, a real-time modeling and visualization system for vehicle emissions on an urban road network (RTVEMVS) was designed and built with Web platform by the development of three modules: real-time data collection, real-time modeling and real-time visualization. The real-time mapping of vehicle emission by vehicle types and road links and real-time“precise vehicles emissions impact estimation” technology which can trace key roads and vehicles induced high air pollution were innovated on system to provide the sensitive perception for government decision makers on speedy vehicles emission control. The detailed vehicle information, traffic flow, speed on each road link and weather data were accessed and fused in real-time from multiple urban sensor systems together with road network information. Integrating an emission calculation model with a high spatiotemporal resolution and the AREMOD dispersion model was running on real-time in RTVEMVS. Based on the Web interface and WebGIS technology (ArcGIS API for JavaScript), real-time vehicle emissions and estimation results can be visualized on a map. The RTEMVS was demonstrated by application on Chongqing of China, hourly based-links vehicle emission by vehicle types on road network and the top 10 of key road links and vehicles induced high air pollution per hour were clearly reveal and visually viewed, and indicate that the real-time system with real-time mapping and precise estimation can provide direct support to the variable vehicle emission control. The RTVEMVS is a useful platform for supporting decision-making in the precise control of urban vehicle emissions pollution.

Fast travel-distance estimation using overhead graph

ABSTRACT Shortest path can be computed in real-time for single navigational instructions. However, in complex optimisation tasks, lots of travel-distances (lengths of shortest paths) are needed and the total workload becomes a burden. We present a fast method for estimating the travel-distance from one location to another by using an overhead graph that stores the ratio between the bird-distance and the travel-distance for few representative locations. The travel-distance is then estimated for any two locations using the corresponding value between their nearest nodes in the graph. We test the method within an optimization setting where the goal is to relocate health services so that the travel-distance of patients is minimised. We build the overhead graph using road network information from Open Street Map and experiment with real-world data in the region of North Karelia, Finland as a part of the ongoing IMPRO project. The results show that the average estimation error is 0.5 km with a graph of 512 nodes, and the total processing time reduces from 1.2 hours to 2.9 seconds per iteration in the optimisation process. The error in the estimated travel-distances is 2%, on average, which is significantly smaller than 8% of the second best estimation method.

Exploring the Spatial Distribution Characteristics and Correlation Factors of Wayfinding Performance on City-Scale Road Networks Based on Massive Trajectory Data

Understanding how urban residents process road network information and conduct wayfinding is important for both individual travel and intelligent transportation. However, most existing research is limited to the heterogeneity of individuals’ expression and perception abilities, and the results based on small samples are weakly representative. This paper proposes a quantitative and population-based evaluation method of wayfinding performance on city-scale road networks based on massive trajectory data. It can accurately compute and visualize the magnitude and spatial distribution differences of drivers’ wayfinding performance levels, which is not achieved by conventional methods based on small samples. In addition, a systematic index set of road network features are constructed for correlation analysis. This is an improvement on the current research, which focuses on the influence of single factors. Finally, taking 20,000 taxi drivers in Beijing as a case study, experimental results show the following: (1) Taxi drivers’ wayfinding performances show a spatial pattern of a high level on arterial road networks and a low level on secondary networks, and they are spatially autocorrelated. (2) The correlation factors of taxi drivers’ wayfinding performances mainly include anchor point, road grade, road importance, road complexity, origin-destination length, and complexity, and each factor has a different influence. (3) The path complexity has a higher correlation with the wayfinding performance level than with the path distance. (4) There is a critical point in the taxi drivers’ wayfinding performances in terms of path distance. When the critical value is exceeded, it is difficult for a driver to find a good route based on personal cognition. This research can provide theoretical and technical support for intelligent driving and wayfinding research.

Extended geographic information monitoring of urban ecotone based on cell transmission model

The road traffic network in the urban interlaced zone is a collection of road sections and nodes. The difficulty in road traffic monitoring lies in the processing of traffic at the node locations. In this paper, the cell transmission model is used to monitor the extended geographical information of the urban interlaced zone. Using the cell transmission model, the channelisation and cell outflow capacity at the intersection are calculated. According to the timing of traffic signals in the interlaced zone, calculations are made on the road traffic volume and accurately monitoring the vehicle density. The simulation experiment results show that the model in this paper can accurately monitor the congestion information of the gradation road network and the uniform road network in the urban interlaced zone traffic network, and the monitoring results are close to the actual results, indicating that the research method in this paper has stronger applicability.

Distribution and Accessibility of Road Networks to Educational Facilities in Delta State: Transport Geographical Appraisal

This study examined the geospatial distribution and accessibility of road networks to the educational facilities in Delta State: A transport geographical appraisal. The administrative map of Delta State was acquired and used to prepare the geographical information of the entire road networks of the study area. The choice of nodes was based on population size. Based on the adopted operational definition of major centres, 50 major centres were identified. The Pearson's product-moment correlation coefficient, as well as the regression techniques, were used in the analysis. A high correlation coefficient value (r = 0.60) was obtained between post-secondary institutions (x1) and primary institutions (x3). For access to have meant it should be people-oriented. The correlation between population and functional index of educational facility occurrence gives (r = 0.13). This means that large population centres tend to have fewer numbers and lower order types of educational facilities in the study area. The coefficient of determination of the relationship is given as 0.017 which indicates that population has no effect on the distribution of educational facilities in Delta State. This implies that population alone does not attract the establishment of educational facilities while small population centres tend to be favoured in the study area. Based on the findings recommendation was proffered.

Road Extraction in Mountainous Regions from High-Resolution Images Based on DSDNet and Terrain Optimization

High-quality road network information plays a vital role in regional economic development, disaster emergency management and land planning. To date, studies have primarily focused on sampling flat urban roads, while fewer have paid attention to road extraction in mountainous regions. Compared with road extraction in flat regions, road extraction in mountainous regions suffers more interference, due to shadows caused by mountains and road-like terrain. Furthermore, there are more practical problems involved when researching an entire region rather than at the sample level. To address the difficulties outlined regarding mountain road extraction, this paper takes Jiuzhaigou county in China as an example and studies road extraction in practical applications. Based on deep learning methods, we used a multistage optimization method to improve the extraction effect. First, we used the contrast limited adaptive histogram equalization (CLAHE) algorithm to attenuate the influence of mountain shadows and improve the quality of the image. Then the road was extracted by the improved DSDNet network. Finally, the terrain constraint method is used to reduce the false detection problem caused by the terrain factor, and after that the final road extraction result is obtained. To evaluate the effect of road extraction comprehensively, we used multiple data sources (i.e., points, raster and OpenStreetMap data) in different evaluation schemes to verify the accuracy of the road extraction results. The accuracy of our method for the three schemes was 0.8631, 0.8558 and 0.8801, which is higher than other methods have obtained. The results show that our method can effectively solve the interference of shadow and terrain encountered in road extraction over mountainous regions, significantly improving the effect of road extraction.

Deep learning enabled vehicle trajectory map‐matching method with advanced spatial–temporal analysis

Global positioning system (GPS) trajectory map matching projects GPS coordinates to the road network. Most existing algorithms focus on the geometric and topological relationships of the road network, while did not make full use of the historical road network information and floating car data. In this study, the authors proposed a deep learning enabled vehicle trajectory map-matching method with advanced spatial–temporal analysis (DST-MM). The algorithm mainly focused on the following three aspects: (i) analyse the spatial relevancy from the prospective of geometric analysis, topology analysis and intersection analysis; (ii) to make full use of the historical and real-time data, a deep learning model was conducted to extract the road network and vehicle trajectory features and (iii) establish a speed prediction model and nest it in the temporal analysis structure. It narrows down the path search range through establishing the dynamic candidate graph. Experimental results show that the proposed DST-MM algorithm outperforms the existing algorithms in terms of matching accuracy for low-sampling frequencies GPS data, especially in the central urban area.

Citywide traffic speed prediction: A geometric deep learning approach

Accurate traffic speed prediction is critical to modern internet of things-based intelligent transportation systems. It serves as the foundation of advanced traffic management systems and travel services. Nonetheless, the large number of roads and sensors impose great computational burden to existing forecast approaches, most of which can only handle one or few roads at a time. In this paper, a novel data-driven deep learning-based approach is proposed for citywide traffic speed prediction. The proposed approach is grounded on recent developments of geometric deep learning techniques to fully utilize the topological information of road networks in the learning process. Specifically, the approach captures the geometric traffic data dependency with graph convolution and attention mechanisms, and the temporal data correlation is extracted and expanded using the encoder–decoder architecture within a generative adversarial learning framework. Comprehensive case studies are conducted with real-world urban road networks and respective data to evaluate its performance, where consistent improvements can be observed over baseline approaches. Lastly, an architectural study is carried out to discover the best-performing structure of the proposed approach, whose sensitivity to data noise and sample frequency is also assessed.