3D RECONSTRUCTION AND RENDERING VISUALIZATION OF TUNNEL POINT CLOUD

Abstract. This paper presents a method for rapid and automatic visualisation of over-undercutting based on 3D reconstruction and rendering of tunnel point clouds. The method uses model filtering and our improved voxel filtering for point cloud lite to maximise the preservation of the original tunnel point cloud features while streamlining the point cloud. The 3D reconstruction is then completed according to our proposed alpha-shape algorithm with low reconstruction parameters combined with a fast model optimisation model, which achieves high reconstruction accuracy while significantly improving computational efficiency to meet the requirements of the tunneling project. The rendering method is calculated by selecting the RGB values. The darker the colour, the heavier the degree of over-excavation or under-excavation. According to the calculated over-excavation or under-excavation value, the ratio of the point in the over-excavation or under-excavation interval is obtained in turn, and the gradient is rendered by increasing the base colour of each interval proportionally; the kd-tree is then introduced to calculate and visualise the colour information of the reconstructed model. In addition, this paper uses real-world tunnel point cloud data and demonstrates that the method meets expectations through qualitative and quantitative evaluation. Visualisation of the tunnel point cloud is accomplished while maintaining speed, visualising the over-under-excavation situation and providing an outlook for future work.