Color Component–Based Road Feature Extraction from Airborne Lidar and Imaging Data Sets

Abstract

AbstractThis paper presents a new framework of road feature extraction from color component–based data fusion of aerial imagery and lidar data. The proposed framework consists of six procedures: (1) removal of elevated objects (e.g., buildings) from lidar data with a flatness index constraint; (2) removal of shadows and vegetation from aerial images using the Otsu segmentation; (3) data fusion of the modified lidar data and aerial images; (4) initial extraction of road features from the fused data; (5) refinement of road features to remove false positives and join up misclosures; and (6) final extraction of road surfaces and centerlines. A new method is proposed for data fusion of aerial images and lidar data to extract road features by utilizing color components, such as luminance, saturation, and hue, in hue/saturation/intensity and brightness/blue difference/red difference color spaces. A series of refinement processes, including hierarchical median filtering and k-nearest-neighborhood, are implemented...