Road Line Extraction Based On DMI

In the smart car driver assistance field,road line detection and tracking extractextraction based on sight is a key technology.Safe driving and intelligent autonomous vehicle navigation of research is of great significance.Many current scholars made significant achievements in this regard,but most are for some special structural road proposed model,and can not be applied to a variety of road conditions,especially when the pavement structure of the road is not being tracke d uniform,strong light,shadow or other sheltered complex road conditions,road line detection,tracking algorithm larger deviation,leading to detection,tracking extraction failed.This study is the use of Street View Street View image acquisition syste m to carry out road extraction lines.Research papers divided into three parts: the road image pre-processing method,a road line detection algorithm,roads tracing algorithm,and ultimately through the experimental verification algorithm availability.Am ong them,the road image pre-processing method includes an image graying,median filtering denoising,image adaptive histogram equalization,edge enhancement,image binarization,morphology repair.Road line detection algorithm is divided into regions of i nterest extraction and fitting a straight line in two parts.Specifically: First,make good use of the region of interest extraction dichotomy road;after the region of interest is determined,using the above image preprocessing algorithm for image process ing;and on this basis,the use of structural elements morphological feature points on the road information store,under certain constraints on the lane line filter information.Finally,Hough transform linear least squares fitting method,this method can greatly improve the recognition accuracy.Path tracing algorithm is mainly used on a road line image information to predict the next frame image path feature region,and then after the image pre-processing and fitting a straight line to complete the final track of the road line extraction task.Through statistical analysis of experimental,the results prove that the proposed Hough Transform and Least Squares Method road line detection of regional roads in the basic line can be extracted,the success rate of over 80%.However,for some shelter,road line detection of defects and other special circumstances remain undetected,false detection situation still requires in-depth study.