Research On Unstructured Road Recognition And Fitting Algorithm

Intelligent vehicle autonomous driving and navigation is a research focus in the field of the world, the smart car in the outdoor environment in accordance with the established objectives and tasks, combined with the known geographic information to make path planning, through continuous perception and judgment of environmental information in the process of autonomous driving around in the direction of travel of the vehicle, automatic adjustment and operation the corresponding control vehicle. The technology of the road recognition is the key technology of intelligent vehicle unmanned systems, good real-time performance, strong adaptability, high robustness of road recognition algorithm is an important guarantee to realize autonomous driving vehicle. This paper focuses on the exploration and Research on unstructured road recognition algorithm and the lane line fitting algorithm.This paper proposes a recognition algorithm based on the gray level features of road, the algorithm first preprocesses the road images, including gray, image segmentation, a block mark method of Road area by use of the image, and then use the improved edge along the regional road extraction region growth method, analysis on both sides of the road, along the edge of curvature, identifying road type and turning direction. The expansion of regional road edge line extraction out of range, marking the road boundary. Finally on the road boundary region with Canny edge detection algorithm to detect road edges. Of Road area, road area along the edge of the edge line expansion area and roads, to segment the road around the lane line, using the method of least squares fitting the lane line. The error of a large number of experiments and data analysis, the algorithm has good recognition effect on a straight line and the road, to meet the basic indicators of the evaluation system of the performance of the algorithm, has a certain practical application value.