Single-View 3D Model Fast Reconstruction Method Based on Mesh Deformation

Object model reconstruction plays a vital role in enabling robotic tasks, such as grasping and human-robot collaboration technology. However, the reconstruction methods based on single-view images lack depth information and suffer from lower quality reconstruction for occluded parts, which limits the model's robustness and generalization capabilities. To address these challenges, this paper proposes a single-view 3D model reconstruction method based on model retrieval and deformation. By constructing a database to store source models and incorporating graph convolution and attention mechanisms in network design, the method enables efficient model selection and surface reconstruction. Additionally, a simulation scene is established for validating the object grasp. This method holds significant potential for applications in human-machine collaborative scene, providing an effective solution for object K-Nearest Neighbor grasp tasks.