Research On The Detection And Evaluation Of Road Traffic Marking Retroreflection Coefficient Based On Vehicle LiDAR

As an important part of road traffic safety,road traffic markings provide safe and reliable guidance for traffic participants.With the passage of time,the reflective effect of road traffic markings will decrease.As an important index to measure the reflective effect of road traffic markings,the coefficient of retroreflected luminance plays a vital role in maintaining road safety.The existing portable retroreflective measuring instrument can only detect the coefficient of retroreflected luminance manually,and the cost is high and the detection efficiency is low.This paper aims to use vehicle LiDAR to obtain the reflection intensity of road traffic markings and predict their coefficient of retroreflected luminance,so as to realize the high efficiency,low cost and automatic detection of the coefficient of retroreflected luminance of road traffic markings.In this paper,the point cloud data of road environment obtained by vehicle LiDAR system and portable retroreflective measuring instrument and the coefficient of retroreflected luminance of road traffic markings are taken as the research objects.By studying the extraction of road traffic markings,the coefficient of retroreflected luminance is predicted by using the reflection intensity of road traffic markings,and the reflection effect of road traffic markings is realized.The evaluation has great practical significance for maintaining road traffic safety.In the process of detecting and evaluating the retro-reflection coefficient of road traffic markings based on vehicle LiDAR,firstly,the coefficient of retroreflected luminance and road environment point cloud data of road traffic markings on the same road section are obtained by using portable inverse reflection measuring instrument and vehicle LiDAR system respectively.For the road environment point cloud,after preprocessing,the slope filtering algorithm and the improved region growing algorithm are used to complete the extraction of the road point cloud data.At the same time,according to the difference of the reflection intensity between the road surface point cloud and the road traffic marking,the road traffic marking is obtained by using the histogram statistical method.Then,the RANSAC algorithm is used to extract and segment the middle four road traffic markings for calibration,and the reflection intensity value in the segment is output.The statistical analysis of the reflection intensity and the coefficient of retroreflected luminance shows that the Pearson correlation coefficient between the coefficient of retroreflected luminance and the reflection intensity information is 0.511,and the significance level is 0.01.Aiming at the coefficient of retroreflected luminance and reflection intensity data,the coefficient of retroreflected luminance is estimated by polynomial regression,multiple linear regression and decision tree regression.According to the analysis of model evaluation index,a prediction model of coefficient of retroreflected luminance of road traffic markings based on decision tree regression is established.Finally,the road traffic markings that need to be maintained are determined by the results of the coefficient of retroreflected luminance of the verification area.Experiments show that the model prediction results are basically consistent with the traditional methods,and the time cost is reduced by at least 80%.It is proved that the retro-reflection coefficient detection of road traffic markings based on LiDAR proposed in this paper can realize the efficient,automatic and global detection of the coefficient of retroreflected luminance of road traffic markings.The proposed decision tree regression model can accurately predict the coefficient of retroreflected luminance of road traffic markings and realize the evaluation of the reflection effect of road traffic markings.