Research on three-dimensional (3D) reconstruction method for municipal drainage pipelines based on sonar

Aiming at the internal inspection challenges of drainage pipelines under fully filled, nearly filled, and turbid water conditions, traditional closed-circuit television fails to image effectively due to poor light transmittance. Existing circular scanning sonar methods are restricted by strong noise interference, difficulty in separating point clouds of pipe walls and bottom sediment, and poor continuity of three-dimensional (3D) reconstruction. This paper proposes a 3D pipeline reconstruction method based on circular scanning sonar, covering single-frame point cloud processing, geometric parameter extraction, and joint 3D reconstruction of pipelines and sediment. In single-frame processing, external noise is removed by a hybrid filtering framework optimized by a genetic algorithm. Then, a two-stage strategy of initial Euclidean clustering and angular density clustering in polar coordinate space is adopted to stably separate pipe wall and sediment point clouds, followed by improved bilateral filtering to suppress internal noise. The pipe wall is fitted with a weighted Gaussian-Helmert model to obtain the center and radius, and the sediment boundary is fitted by a method combining first-order interpolation and quadratic spline correction to avoid local distortion. In multi-frame reconstruction, the pipeline central axis is estimated by smooth fitting of the center sequence, and the cross sections are uniformly calibrated. The pipe wall is reconstructed by axial interpolation and lofting, and the sediment boundary is reconstructed by normalized parameterization and axial continuity constraints. Finally, a joint 3D model of the pipeline and sediment is formed, which can output sediment height, area, and volume parameters along the axial direction. Experiments are conducted on data collected from a simulated site and a real drainage network in Chengdu. Compared with other methods, the results show that the proposed method meets the requirements in terms of geometric error of the pipe wall, axial continuity, and stability of sediment parameters. It can effectively restore the real 3D morphology of pipelines and provide technical support for water-filled pipeline inspection and quantitative sediment analysis.