PointRecon: Online 3D Point Cloud Reconstruction via Ray-based 2D-3D Matching

Chen Ziwen, Zexiang Xu, Fuxin Li

Keywords: 3D Reconstruction

Abstract:

We propose a novel online point-based 3D reconstruction method from a posed monocular RGB video. Our model maintains a global point cloud scene representation but allows points to adjust their 3D locations along the camera rays they were initially observed. When a new RGB image is inputted, the model adjusts the location of the existing points, expands the point cloud with newly observed points, and removes redundant points. These flexible updates are achieved through our novel ray-based 2D-3D matching technique. Our point-based representation does not require a pre-defined voxel size and can adapt to any resolution. A unified global representation also ensures consistency from different views. Results on the ScanNet dataset show that we improve over previous online methods and match the state-of-the-art performance with other types of approaches.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 5169