Research On Road Detection Technique Based On Vision Navigation Of Intelligent Vehhcle

Recently, road detection technique is a very important part of vision navigation, which is the key technique of intelligent vehicle guidance. The detection result is seriously affected by the quality of noise and image. Several key problems in the road detection technique are chosen as the emphasis of this paper, and then modify the road image quality by a series of changes in image enhancement, edge detection and so on. This paper analyses and researches the methods of road detection technique based image process.This paper analysis the type of noise, put forward the right denoise method for different types of noise, and then proposed an improvement in histogram equalization, which can effectively solve the blurring problems in the image. For certain degree influence regarding the path recognition under the sleet condition to use spiced salt noise to carry on the simulation to it. The discovery uses automatic filter the effect to white Gaussian noise is better.Through the specific analysis and comparison of various edge detection algorithm, the image edge detection algorithm based on multiscale geometry analyse is used to extract the edge outline. The new edge detection algorithm can make use of the geometrical regularity better and conform to best practices in image picture attribution. By the simulate experiments, the method processes lubricity and series in the edge, the texture of road edge is protected better. In image segmentation, this paper study server traditional threshold segmentation algorithm, and use maximum classes square error in this paper.In road boundary line extraction, use Hough transform based on Line Point Density, which can solve the accumulating problem of classical Hough transform. This method also have nice anti noise performance and robustness, and can effectively extract lines form road image. Add spiced salt noise, white Gaussian noise and fuzzy treatment methods in the real-time simulation of the road likely to be encountered in the quality of the image, then pretreatment for road detection. The proof shows that the algorithm in this paper can effectively improve the recognition rate in road detection, accurately extract lines form road image, and proved the efficiency and the reliability of this approach. It also provided useful road information for follow up study.