Lane detection via object positioning systems based on CCD array geometry

This paper presents an approach to lane detection for a vehicle. The positions of the lane marks can be evaluated by visual information of the image captured from a single charge-coupled device (CCD) camera. This proposed approach originally utilizes the properties of the CCD array in a camera to achieve the aim of objects positioning, since one pixel information produces two equations and increases one unknown variable. After camera calibration, this approach can therefore evaluate the intrinsic parameters of a camera from more pixel information. The configuration of the CCD chip cells is the key factor in this approach. The pixels of a resulting image directly reflect the geometry of the CCD cell, or the CCD array, in the camera. According to the attitude of the camera, this paper constructs the coordinate transformation that can resolve the geometrical relations between the film coordinate (the CCD array) and a fixed coordinate. This paper also provides associated techniques to facilitate the proposed approach, including image geometry analysis, distribution analysis of the CCD array, least mean square error (LMS) algorithm, etc. A down scaled experiment for lane detection is used to verify the feasibility of the proposed approach. The results show that the proposed approach is able to achieve object positioning.