| 摘要 |
The present invention relates to video-based driver assistance systems being used in vehicles today to perform a variety of functions, such as lane or road departure warning. These functions often require that parts of the video image be identified as, for example, a traffic sign, a bridge, lane markings, etc. Particularly the present invention relates to a method and system for online calibration of a vehicle video system using vanishing points, evaluated from frames of a camera image contains identified markings or edges on a road. The vanishing points are located by finding or extrapolating at least at one left- and/or right-hand side markings or edges of a road to a point of intersection, whereby a long term average vanishing point location is calculated with time-filtering methods from a sequence of images, even when only one side of a road or lane marking or an edge is visible in any given frame of an image at a time, and from the time-averaged vanishing point location coordinates, the static yaw and pitch angles of the camera are deduced. The present invention relates to also to a method to locate the vanishing points by finding or extrapolating the regularity of repeated road markings, such as dashed patterns and to a method to locate the vanishing point d by using the change in spatial frequency of textures on the surface ahead as an object to determine the camera pitch angle. |
