A Novel Lane Model for Lane Boundary Detection

This paper addresses the problem of the navigation of a mobile vehicle in outdoor environment. A CatmullRom Spline based lane model which described the perspective effect of parallel lines was proposed for generic lane boundary. As Catmull-Rom Spline can form arbitrary shapes by control points, it can describe a wider range of lane structures than other lane models such as straight and parabolic model. It formulates the lane detection problem in the form of determining the set of lane model control points. The proposed algorithm uses a Maximum likelihood approach to the lane detection problem. In contrast to other approaches like Edge-Link and Snake, the proposed algorithm can increase the accuracy of estimation which approximates to global optimum. The proposed method is robust to the noise present in the road image with shadows, variations in illumination, marked and unmarked road. In Section 2, reviews on existing lane detection techniques are presented. Section 3 presents a new lane model based on the Catmull-Rom spline. In Section 4, a proposed algorithm which uses the new lane model and the maximum likelihood method for lane detection is developed. Section 5 shows representative results of applying this proposed lane model and algorithm to various road types and environments. Conclusions are given in Section 6. At present many different vision-base lane detection algorithms have been developed. They depend on different road models (2D or 3D, straight or curve) and approach techniques (Hough, template matching, neural networks, etc.). A curve road model is proposed by [1][2][3], it