Combining Images and Trajectories Data to Automatically Generate Road Networks

Road network data are an important part of many applications, e.g., intelligent transportation and urban planning. At present, most the approaches to road generation dominated by single sources including images, point cloud data, trajectories, etc., which may cause fragmentation information. This study proposes a novel strategy obtain vector networks combining images trajectory with postprocessing method named RNITP. The designed RNITP includes two parts: initial layer detection map acquirement. first layer, there three steps information interpretation from based on new deep learning model (denoted as SPBAM-LinkNet), trajectories rasterizing, fusion using OR operation. last is used generate that focused error identification removal. Experiments were conducted kinds datasets: CHN6-CUG datasets HB datasets. results show accuracy, F1 score, MIoU SPBAM-LinkNet (0.9695, 0.7369, 0.7760) (0.9387, 0.7257, 0.7514), respectively, better than other typical models (e.g., Unet, DeepLabv3+, D-Linknet, NL-Linknet). In addition, IoU, recall obtained 0.8883, 0.7991, 0.9065, respectively.