Automated Highway Sign Extraction Using Lidar Data

1 Traffic signs are an integral element of any transportation network; however, keeping records of 2 those signs and their conditions is a tedious process, which is both time consuming and labour 3 intensive. As a result, many agencies around the world have been working towards automating the 4 process of sign collection. One form of automation is using remote sensing techniques to extract 5 traffic sign information. This paper proposes an algorithm, which can be used to automatically 6 extract traffic signs from mobile Light Detection and Ranging (LiDAR) data. The number of signs 7 on a road segment is identified and the coordinates of those signs are used to map the data onto 8 the road segment. The sign extraction procedure involves applying multiple filters to the point 9 cloud data and clustering the data into traffic signs. The proposed algorithm is tested on three 10 different highways located at different regions around the province of Alberta, Canada. The 11 segments on which the algorithm was tested included two-lane undivided rural road and four-lane 12 divided highways. Geometry on the different highways varied as did the vegetation and tree 13 density. Success rates ranged from 93-to-100% with the algorithm performing better on highways 14 where overhead signs did not exist. The results show that the proposed method is simple yet 15 effective in collecting inventories of traffic signs at high precision rates. 16 17