Overall flowchart of online measurement method for complex pipelines based on area pre-segmentation

(Image by the research group)

Complex tubes with flanges, joints and other components, characterized by a large number, diverse shapes and complex structures, are key parts of aircraft engines, rockets and other quipment. In order to ensure the smooth operation of these equipment, the dimensional accuracy of the tube is must be ensured.

Recently, a research team from the Shenyang Institute of Automation (SIA), Chinese Academy of Sciences (CAS) , proposed a complex tube parameters based on the area pre-segmentation of the multi-camera vision of the online measurement method, which can be used to weld the spatial dimensions of the flanged, jointed tube online inspection.

The research team established a complex tube segmentation model based on Triple Detail Feature Capture Fusion (TDC). The model achieves the fusion of multi-scale features through the TDC module, enhancing the ability to integrate contextual information and facilitate inter-dimensional information interaction, allowing the model to focus more on the flange and joint areas, thus achieving precise segmentation of these components.

For the segmented flange region, the research team proposed a flange parameter measurement algorithm based on the edge of the upper surface of the flange. The edge information of the upper surface of the flange is extracted by analyzing the curvature change of the edge of the flange, and the measurement of the flange parameters is realized by 3D reconstruction of the upper surface of the flange. For the segmented tube body region, the research team completed the accurate measurement of tube body parameters by extracting the tube body region that is mutilated in a linear structure and reconstructing the center line of the region. The results show that the method can complete the measurement of a tube in 7 seconds, the flange parameters measurement accuracy is 0.17mm, the tube body parameters measurement accuracy is 0.16mm.

This research is published in the Journal of Manufacturing Processes under the title A multi-camera vision online measurement method for complex tube parameters based on regional pre-segmentation. This study features SIA Ph.D. candidate Hao Zhao as the first author, and Researcher Renbo Xia and Researcher Tao Zhang of SIA as the corresponding authors.

DOI：10.1016/j.jmapro.2025.05.083

Link to paper：https://www.sciencedirect.com/science/article/pii/S1526612525006553?via%3Dihub