The Research About Traffic Signs Line Identification And Early Warning In The Road

With the improvement of people's econo mic level,car ownership increases every year,people travel more conveniently,along with these convenient,traffic congestion and traffic accident are also increasing.In recent years,security driver assistance,intelligent vehicles receiving more and more attention,this product can bring the gospel for people by eff-ectively reduce or even avoid traffic accidents.Among thee,the recognition and warning of the road traffic sign lines as an important branch of the auxiliary driving safety,in which plays a pivotal role.This is the content of this article.Aiming above,extracted from the pavement area,lane identification and warning and Identification and tips of the indicating signs on the road are studied.Mainly in the following work:(1).Due to the lane lines and traffic signs are objects of this paper,and these target oobjects are located on the road,the exact area to extract the road becomes a major task.In this paper,the method based on region growing have a series of improvements,and ultimately proposed a search method of the load surface which combined the dynamic multi-block area of growth and the characteristics of the YcrCb pixel.Firstly,,the approximate range is searched.Then,analysis and exclude the outline information of the image by eliminating obvious non-road area.Ultimately,the pavement area is posited precisely,and pave the way for subsequent extraction of road traffic signs lines.(2).Use the global search method based on HSV and RGB color information to extract all clicks,and the point set is labeled to analysis the connectivity domain,obtaining the maximum width of each connected region as the width of the area Unicom,setting the lower and upper threshold,the non-lane area were initially ruled out;Using the method of least squares fitting lane to fitting point after the points which are suspected to t.he lane are obtained,vanishing points based on the principle of exclusion,after fitting non-lane information,and ultimately will identify for the yellow lane dividing line,a single lane,the work of identifying the actual situation of the three lines.(3).An adaptive lateral calibration method is implemented,the image pixel width between the two lanes on the image according to the position y,and then derive the distance between the vehicle and the lane marking based on the relationship between the current pixel width and the actual width.The vehicles are divided into non-deflection and deflection state,in different running state,the calculation formula is different,the different warning levels of a yellow solid and the solid white line under predetermined conditions are achieved.(4).In order to identify the class of traffic signs on the road indicating the line,a more comprehensive sample information library is established in this paper,the thought that the current flag is positioned by build trapezoidal area between lanes is used,and a standardized operation of the shape and size of the flag is made to achieve the Hu moments and Zernike m oment feature extraction feature extraction both methods,and ultimately by the similarity measure to do the minimum Euclidean distance classifier for classification,combined with the color of the current lane,linear corresponding prolmpt information to the driver.In this paper,the recognition and departure warning judgment of lane and the identification and tip of the traffic signs line are studied under VC++6.0 and OpenC V platform.experiments show that Oour algorithm can better accomplish lane and instruct the class mark line identification,and the combination of the two corresponding warning to the driver or tips.