Research On Tunnel Surface Reconstruction Based On Laser Scan Data

With the rapid development of urban underground rail transit, the safety of subway tunnel has become a widespread concern, therefore the deformation monitoring of tunnel has received more and more attention. Laser scanning technology has been widely used for its characteristics of high precision, high density, real-time and initiative.The use of laser scanning technology in subway tunnel deformation monitoring is a totally new approach in this area. This is a new detecting technique, It can make up the deficiencies of traditional methods and it can make full access to structure information with its non-contact, real-time online mode; This technique can realize multi-sensor data fusion based on computer network; It can improve the accuracy and the ability of overall monitoring in space and time; The technology also avoids the conflict between monitoring and operation, it can avoid the poor working conditions of traditional detecting methods.As the laser scan data has a large number of gross errors, it is a great challenge to the data processing. This paper analyzes the characteristics of the tunnel's laser scan data error,we point out that the laser scanning data errors exists in the radial direction and they are not simply exist in the vertical direction. For such errors, we propose using the method of space conversion to transform the rectangular coordinate space into the polar coordinate space, by doing so the direction of errors can be resumed. As the tunnel's laser scan data contains a great number of gross errors, we must resort to robust estimation methods. In this paper, we have introduced the robust estimation theory and a variety of robust estimation methods. We prove that the method of sign-constrained robust least squares is reasonable and we have improved the algorithm of this method.The findings of this paper are summarized as follows:(1) By using the M-estimation model of Laplace distribution, we have proved that the method of sign constrained robust least squares is reasonable.(2) For the shape of the tunnel, we have improved the algorithm of the sign-constrained robust least squares and greatly improved its efficiency.(3) We have analyzed the error direction of the tunnel laser scan data and propose the method of space conversion to resume the error direction of the tunnel's laser scan data. For the problem of outliers, we successfully removed it by using the improved sign-constrained robust least squares.