Abstract:The noise in the point cloud data of a tunnel affects the accuracy of analysis. Effectively removing the noise data has become the key factor in point cloud based tunnel deformation analysis. A tunnel point cloud denoising algorithm based on centerline was proposed in this paper. First, the gesture of the point cloud was obtained by projecting it onto the horizontal plane and vertical plane respectively, and fitting two curves by high order polynomial equations from the planar data out of which the control points of the centerline were then interpolated. The control points were are used to express the centerline were densified according to the intersection angle between the spatial lines. Meanwhile, the tangent planes at each control point were computed to segment the tunnel point cloud. In addition, the distances from each point to the centerline in each block were computed to compare with the given distance threshold to accomplish data denoising. The feasibility and accuracy of the proposed method were demonstrated by two experiments. The first test is to fit the centerline by simulating the tunnel point cloud while the fitting accuracy of the centerline was analyzed by comparing it with the given one. The second test was to analyze the practical tunnel point cloud. Data denoising was finally achieved by implementing the proposed method.
