Computing vanishing points for self-steering vehicles using video image features
Nic P. Snailum · AIP conference proceedings · 2000
This paper describes a method of finding the vanishing point from lines of perspective in a video image in order to derive self-steering information for a mobile robot. Finding the vanishing point reliably in a cluttered video image can be computationally intensive and often suffers from the problem of accumulations split across several accumulator bins. A polar histogram search method is investigated which is also found to suffer from this problem, and a solution is presented which is then applied to real imagery obtained from an autonomous mobile robot.