01555716

Component

The advancement of airborne and terrestrial laser scanning technology has created a revolution in collecting detailed and accurate three-dimensional (3D) surface information by cutting the data collection time drastically as compared to conventional techniques. However, the mostly manual feature extraction has increased the processing time to more than conventional techniques. In order to overcome that, the paper has demonstrated a comprehensive methodology to automatically extract road features with no manual intervention. Based on the availability of one or more data sources, the discussed methodology can be used for road feature extraction. In order to simplify the steps, the methodology is broken down into four major steps, namely: data preparation, road localization, feature classification and extraction. Several challenges needed to be overcome with innovative approaches. For example, the challenge of identifying point-to-point correspondence between multi-modal data for co- registration was solved by adopting feature-based registration methods. A RANdom SAmple Consencus RANSAC based vector classification technique was used to create a bare-earth surface model. Furthermore, another classification method, mean shift, which has a capability to be used with both vector and raster data, was adapted to segment road, sidewalk and median. Finally, by a spatial overlay, segmented regions and planes were integrated. The integration model was used to extract 3D model of various road features. In addition to road, sidewalk and median, there are other structures visible in the corridor, such as lamp posts, power lines and posts, on and off road signs, etc., which will be of interest in the future extraction research.

This paper was sponsored by TRB committee AFH30 Application of Emerging Technologies to Design and Construction.

01550057

15-3394

English

19p

2015

Transportation Research Board 94th Annual Meeting

Digital/other

Figures; References; Tables

Classification; Data collection; Data fusion; Infrared detectors; Infrastructure; Lasers; Mathematical models; Roads; Roadside structures

Laser scanners

Data and Information Technology; Design; Highways; I20: Design and Planning of Transport Infrastructure

Transportation Research Board Annual Meeting 2015 Paper #15-3394

TRIS, TRB, ATRI

Dec 30 2014 1:08PM