| The traditional digital image technology measures the mean texture depth of asphalt pavement,usually by obtaining pavement gray data to calculate the average gray difference,using a certain proportional correction coefficient to convert the mean texture depth,without analyzing the relationship between gray difference and elevation difference,and the proportional correction coefficient is empirical value,which is not applicable to different surface construction states measured under different conditions.Therefore,by exploring the optimal lighting method of gray level consistent with elevation change,this paper uses laser to obtain pavement elevation data to check image grayscale,studies the regression model between gray difference and elevation difference,establishes a mean texture depth algorithm based on regression model,and explores the influence of various factors on the accuracy of the algorithm.At the same time,based on digital image correlation and DIC technology,this paper establishes a three-dimensional visual reconstruction detection system for asphalt pavement and compares and analyzes it.Firstly,according to the reflection characteristics of asphalt pavement surface,three lighting rendering methods are designed,asphalt mixture models with different surface states are established through blender,and the gray level of rutted model changes with elevation under different lighting rendering methods is compared and analyzed,and the optimal lighting illumination method is determined.Secondly,the corresponding elevation and gray level data of the model cross-section points are obtained,the original gray level is corrected by the correction coefficient,the regression model of the elevation difference-gray difference of the cross-section point is analyzed by the neural network,and the average depth algorithm based on the regression model is established to realize the conversion of the average gray difference to the average height difference,and the conversion accuracy of the algorithm’s gray difference-elevation difference is evaluated by using the average cross-section depth of the section at different positions of the model.Then,according to the 3D point cloud data on the surface of the rutted plate model,the depth matrix is established,the upper surface of the subregion division interpolation fitting simulation method extracts the mean texture depth to calculate the datum,calculates the real mean texture depth of the model through area integration,analyzes the influence of experimental parameters such as threshold,light brightness and incidence angle on the mean texture depth,and determines the best experimental parameters by comparing with the real mean texture depth.Then,the close-up photogrammetry and DIC technology are used to build a binocular visual inspection system,the camera calibration is used to solve the internal azimuth parameters,the external azimuth parameters of the camera are solved according to the relative orientation of continuous image pairs and beam adjustment method,the feature point matching of binocular images is performed by grayscale DIC analysis,the three-dimensional point cloud of pavement is solved based on the principle of binocular vision,and the sub-region division interpolation fitting method and integral calculation are used to mean texture depth.Finally,the laser check image grayscale algorithm,digital image correlation 3D reconstruction method and sand laying method are compared through indoor and outdoor experiments.The experimental results show that the asphalt pavement construction depth algorithm and digital image correlation three-dimensional reconstruction algorithm of laser check image grayscale have good correlation with the sand paving method.The measurement accuracy of the laser check image grayscale algorithm is better than that of the digital image correlation 3D reconstruction method,and the average relative errors of the field test and sand laying method measurement results are 3.69% and 6.41%,and the average absolute error is 0.04 mm and 0.07 mm,respectively.The asphalt pavement mean texture depth algorithm of laser check image grayscale has good stability and high accuracy,which has high engineering practical significance. |
