Paper

Road Network Identification by means of the Hough Transform with Uncertainty Analysis

Volume Number:
10
Issue Number:
1
Pages:
Starting page
58
–
Ending page
72
Publication Date:
Publication Date
1 June 2015
Author(s)
Eric Salerno, Nagavenkat Adurthi, Tarunraj Singh, Puneet Singla, Adnan Bubalo, Maria Cornacchia, Mark Alford, Eric Jones

paper Menu

Abstract

The focus of this paper is on the use of ground target kinematics to estimate the underlying road network on which the vehicles are assumed to be travelling. Assuming that the road network can be represented as an amalgamation of straight line segments, a Hough transform approach is used to identify portion of road which correspond to straight line segments. Since multiple tracks can be associated with one segment of the road and since the track estimates are inherently uncertain, an iterative approach is presented to identify a parametric representation of the line segments of the roads using the total least squares cost function. Cramer´-Rao bounds are identified to characterize the bounds on the uncertainty associated with the proposed approach. A complex dataset which include multiple tracks is used to illustrate the ability of the proposed algorithm to identify the underlying road network and characterize the uncertainty associated with the parametric estimate of the road.