A Guided Deep Learning Approach for Joint Road Extraction and Intersection Detection From RS Images and Taxi Trajectories

Abstract

Automatic extraction of road information based on data-driven methods is significant for various practical applications. Remote sensing (RS) images and GPS trajectories are two available data sources that can describe roads from a complementary perspective, and fusing them together can improve road detection performance. However, existing studies on the combination of RS images and GPS trajectories do not fully utilize their enhanced information about roads and suffer from road information loss. Moreover, roads and intersections are two crucial elements of road network generation, and they are closely related to each other. Therefore, we propose a multi-task and multi-source adaptive fusion (MTMSAF) network, which leverages RS images and trajectory datasets to execute road extraction tasks and intersection detection tasks simultaneously. The MTMSAF network is built on three components. First, two encoder steams are designed to capture road features from RS images and trajectories. Then, an adaptive fusion module is created to fuse the individual road features at each level in a guiding fashion. Finally, two specific decoders are proposed: one is used to implement the road region detection task by considering multidirectional geometric road information, and the other is used to perform the intersection extraction task by recovering intersection information with the road region feature. Four stateof-the-art deep learning-based methods, including segmentation networks and road extraction networks, are compared with the proposed approach. The results show the superiority of our approach for both road detection and intersection extraction tasks.

A Survey of Deep Learning Road Extraction Algorithms Using High-Resolution Remote Sensing Images

“…Roads and intersections are two crucial elements in road network generation. Li et al [ 102 ] using trajectory data and remote sensing images, and not only extracted road surfaces but also recovered intersection information from road area features, simultaneously performing road surface and intersection extraction tasks. Additionally, some researchers apply multi-tasking to segmentation and change detection.…”

Road Intersection Extraction Based on Low-Frequency Vehicle Trajectory Data

“…Hu et al rasterized the trajectory data with different resolutions, filtered and refined by mathematical morphology, and fused the information on intersections extracted with different resolutions according to the rules to obtain the best road intersection results [39]. Moreover, [20,22,[40][41][42][43][44][45]: (1) taking the density peak clustering (CFDP) of trajectory data in the vector space and obtaining intersections on the basis of a mathematical morphology in a raster space and (2) establishing a fusion mechanism to finally obtain road intersection information. Multilevel fusion is also performed on the basis of two data sources, vehicle trajectory and remote sensing images, in the extraction of seed intersections, collaborative training, and integrated recognition to extract the vehicle trajectory intersection features and remote sensing images to identify road intersections.…”

Road intersection detection based on YOLOv5s and topological analysis: a case study of Shijiazhuang

“…Traditional methods of surveying the road network are becoming increasingly unsuitable for the rapid construction and renewal of urban roads, due to the time consuming, low coverage and high cost involved. Remote sensing data are increasingly used by scholars to extract road networks due to their large scale, low cost and high temporal-spatial resolution [2][3] .…”

“…In recent years, many scholars have used deep learning models to extract roads and intersections from remote sensing images, and achieved good results [7] . Li et al proposed a new deep learning framework for multi-tasking, using remote sensing images and vehicle trajectory data to simultaneously perform road extraction and intersection detection tasks, so as to improve road intersections performance [2] . However, the current research is only limited to extracting road intersections, and few studies have constructed their topological relationships and analyzed the spatial distribution of road networks in cities through their topological relationships.…”