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

This 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 to remove open areas (e.g., parking lots) of the road extraction results. A local spatial interpolation method is applied to join up misclosures, and curve fitting is used to obtain accurate road centerlines. The results of tests on sample data sets indicate that the proposed framework performs well, with high accuracy, completeness, and quality.