Research On Tunnel Feature Extraction Based On Laser Point Cloud Data

The 3D laser scanning equipment with high precision and fast detection has been applied graduallyto thetunnel in operation.Currently,the laser scanning data of the tunnel is used to detect the convergent deformation of the section.The section is usually obtained by the model or the method of fitting the central axis.A few of these methods are related to the detection of water leakage disease.The method converts point cloud reflection intensity data into image data,and then uses image processing technology to identify leakage information.However,the of this kind of image is not only related to the selection of color conversion function,but also to the deviation between laser reflection intensity value and actual value,finally the conversion of uncorrected intensity data into images increases the bias further.In this paper,the laser reflection intensity correction algorithm and the method of extracting the central axis of the tunnel are analyzed,and the obtained conclusions are applied to the identification of tunnel leakage and the measurement of section convergent deformation.There are the main tasks of this paper:Based on the experimental data of reflection intensity,the law of reflection intensity distribution of laser scanning data was studied.Combined with the method of multivariate analysis of variance,the significant factors(scanning distance)affecting the intensity value are determined preliminarily.The relationship between this factor and intensity is further determined by single factor analysis.The Taylor's polynomial intensity correction model is proposed.The intensity-corrected point clouds are used to generate the gray image of the inner surface of the tunnel.The accuracy of water leakage disease identification based on those gray image generated of about 100 m and 1000 m tunnels respectively reached 92.3% and 92.8%.By comparing with the results of strength uncorrected and directly using of real image recognition,it's shown that the laser point clouds can quickly and accurately identify the tunnel leakage information by using the model models mentioned in this paper.The cross sections of subway tunnel are extracted and the convergence diameter of tunnel is calculated based on Laser scanning data.This paper presents a novel method for the extraction and cross-section determination of the central axis of the tunnel.This method first acquired the initial central axis points based on the projection slice method,used MST to restore the topological relation of all axis points,used Filter denoising based on least square polynomial smoothing algorithm,segmented the central axis segment and nonlinear segment features,and combined the direction vector of each point on the central axis with the local cylinder fitting model to determine the tunnel section contour points and calculate the diameter of the obtained section.Compared with the results of manual measurement and fixed-point measurement of reflector in this paper,and verifies that the error of fixed-point measurement is millimeter.Based on this method,the diameter error of tunnel section is skewed normal distribution and ?=-14.01569 mm.In addition to the above applications,more research directions and ideas also are extended in the paper,such as 3D reconstruction based on point cloud data,tunnel clearance analysis based on optimization center axis combined with software and tunnel feature recognition method based on modified reflection intensity and spatial distribution law of point cloud,etc.Laser point cloud data can be used in the field of tunnel's disease identification and detection.Based on the intensity correction model,the distance effect of laser reflection intensity can be effectively eliminated,and the classification accuracy of tunnel features and the identification accuracy of water leakage in the generated image can be improved.