Research on pose and position estimation based on mask RCNN edge extraction for AR auxiliary assembly system

Teng Zhu, Xiaoyin Guo, Hansheng Yan, Yongbin Chen, Bin Shu, Yingfa Lu, Changbo Cheng
2022 Third International Conference on Computer Science and Communication Technology (ICCSCT 2022)  
To solve the registration problem of complex texture objects in monocular images in an AR auxiliary assembly system, a 3D object registration method based on Mask RCNN edge extraction is proposed in this paper. This method mainly adopts the edge contour feature of long plate workpiece and enhances the robustness of the algorithm by constructing the matching relationship between 3D model and 2D feature. The experimental result shows that compared with the contour template matching of the Holcon,
more » ... the Mask CNN-based pose estimation method proposed in this paper can effectively improve the effectiveness of the auxiliary system.
doi:10.1117/12.2661820 fatcat:pexjmft4hrdvpkm442phtaee6u