基于多视角图像的关节物体三维重建

In real life, articulated objects are common, such as laptops and drawers. To import these objects into the virtual scene, professional modelers usually need to complete the design and modeling manually. This process is not only time-consuming but also requires a lot of manpower. Algorithms for 3D reconstruction of articulated objects can help the development of fields such as virtual reality and animation industries. However, past algorithms have problems such as high data collection requirements or poor performance under limited viewing angles. In order to overcome these problems, this paper proposes a new 3D reconstruction method of articulated objects based on multi-view images, which can effectively solve the problem of poor reconstruction results in a small number of views. This paper introduces a regularization method to improve the reconstruction effect of the algorithm under limited viewing angle conditions. Through experimental verification, it not only demonstrates the effectiveness of this method in few-view reconstruction, but also proves its potential in improving the quality of 3D reconstruction.This article also proposes anarticulated objects dataset, which can be used for 3D reconstruction of articulated objects.