In submarine pipeline repair projects, the measurement of the deformation characteristics of submarine pipelines is a key link in the repair process. Currently, most of the commonly used measurement methods are difficult to construct a 3D model of the pipeline and accurately obtain the deformation characteristic parameters of the pipeline, and they have the limitation of carrying out construction in complex sea areas. This paper addressed the 3D reconstruction problem of deformed submarine pipelines in reverse modeling engineering and proposed an algorithm for directly extracting the characteristic parameters of deformed submarine pipelines from scattered measurement data. The algorithm was based on the triangulation principle of point cloud data. It extracted feature value data from the deformed submarine pipeline data, used the feature values to fit the cross-sectional contour lines of the submarine pipeline surface, and performed noise reduction through data screening methods to obtain accurate cross-sectional contours and their centers, thereby determining the axis coordinates of the pipeline. In the model reconstruction, the swept sampling method was utilized to uniformly sweep the cross-sectional contour of the pipeline at a fixed step size, and then the overall contour was subjected to Gaussian smoothing processing to obtain a smooth and uniform pipeline model. This algorithm was applied to the deformation repair project of the submarine pipeline in Cezhen, efficiently and accurately extracting the deformation angle, orientation, and other information of the deformed submarine pipeline, providing data support for the pipeline repair construction.