基于蛇模型的管路三维重建方法

Aiming at the difficult problem of high-precision and fast pipeline measurement applied in aerospace, a 3D pipeline reconstruction method was proposed based on snake model. Making full use of the edge features of the gray image of the pipeline and eliminating image noise by means of threshold segmentation and Gaussian filtering, the method was arranged firstly to combine the snake model to obtain accurate pipeline image edges, carry out iteration with the greedy algorithm to ensure convergence efficiency, and use the principle of Epipolar constraint to obtain a two-dimensional pipeline centerline point cloud. And then, a three-dimensional model of pipeline was reconstructed based on the principle of visual projection. Finally, the designed results were optimized to ensure the reconstruction accuracy according to the photos taken by the multi-eye vision system. The results show that the measurement accuracy of the method for pipelines can reach ±0.15 mm, the measurement process is simple and efficient without more operators. The new method can measure the bending radius of the pipeline arc section, adapting industrial applications.