An Integrated Approach to Recognition of Lane Marking and Road Boundary

Abstract

An integrated vision method was proposed for intelligent vehicles to synchronously recognize the lane marking and road boundary in direct or curve road environment. Firstly, with region connectivity analyzing, the method extracted the brightness features of lane markings on input images by self-adaptive threshold segmenting. Not only the gradient magnitude but also the gradient direction features of the road boundary were extracted by the Sobel operator method. Secondly, the 2-D models of road shape were acquired and the features above were matched to them by least-squares fit. With the circular calling of detecting and tracking program block, the whole process showed a fast and exact capability. The experiments have been conducted with the videos captured from real road scenes, and the results proved that it is a real time and robust method to recognize the road for the vision-based navigation of intelligent vehicles.