Road Information Extraction from High-Resolution Remote Sensing Images Based on Road Reconstruction

• DOI: 10.3390/rs11010079
• Publication Date: 2019-01-04T16:34:26Z
• Authors: Tingting Zhou; Chenglin Sun; Haoyang Fu
• Keywords: Tracking (education); Feature (linguistics); Road surface

Traditional road extraction algorithms, which focus on improving the accuracy of surfaces, cannot overcome interference shelter caused by vegetation, buildings, and shadows. In this paper, we extract roads via centerline extraction, width broken connection, reconstruction. We use a multiscale segmentation algorithm to segment images, feature get initial road. The fast marching method (FMM) is employed obtain boundary distance field source field, branch backing-tracking used acquire centerline. Road each calculated combining fields, before tensor applied for connecting gain final matched with its when reconstructed. Three experimental results show that proposed improves solves problem centerline, reconstructing excellent maintain their integrity.