Spherical Harmonics for Robust Next-Best-View Estimation
As 3D depth cameras become more affordable, the additional depth information proves useful in industrial setups. However, with a multiview camera configuration, the complete coverage of a 3D model can only be accurately captured from several views. For this task, the next-best-view (NBV) estimation...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11017667/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849424125287727104 |
|---|---|
| author | Alexandru Pop Levente Tamas |
| author_facet | Alexandru Pop Levente Tamas |
| author_sort | Alexandru Pop |
| collection | DOAJ |
| description | As 3D depth cameras become more affordable, the additional depth information proves useful in industrial setups. However, with a multiview camera configuration, the complete coverage of a 3D model can only be accurately captured from several views. For this task, the next-best-view (NBV) estimation is essential to minimize the actions required for a complete reconstruction. Traditional NBV methods for 3D data are often sensitive to noise and lack rotation invariance. Learning-based methods can make estimations based on current camera information, significantly improving the accuracy of a task. These methods are generally limited to one step, as multi-step planning requires estimating sensor observations. A method that can predict and simulate multiple steps can significantly improve performance when a single step is insufficient. We show that a convex hull of the object can be used to predict n-steps of NBV in a coverage reconstruction setting. The convex hull is used as a proxy for the real geometry of the object in selecting views and requires fewer points sampled from the ground-truth geometry. We propose a volumetric information gain (VIG) score that measures the similarity between the convex hulls obtained from the partial point cloud from selected views and the ground truth point cloud. We retrain an NBV network to predict the views that lead to a point cloud with the highest VIG score, and demonstrate that spherical harmonic preprocessing can mitigate the effect of noise, thereby making the network faster. The convex hull is afterwards used to simulate views, and a set covering optimisation is applied to make a one-shot selection of the best views for surface coverage reconstruction. This method is robust to changes in view space, as shown in the tests applied on Shapenet models along with more complex synthetic shapes from HomebrewDB, Stanford 3D Scanning Repository, and Linemod. |
| format | Article |
| id | doaj-art-04e9bf5dbb9f4fe5bffa6d45871c707a |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-04e9bf5dbb9f4fe5bffa6d45871c707a2025-08-20T03:30:20ZengIEEEIEEE Access2169-35362025-01-011311304911306510.1109/ACCESS.2025.357480411017667Spherical Harmonics for Robust Next-Best-View EstimationAlexandru Pop0Levente Tamas1https://orcid.org/0000-0002-8583-8296Technical University of Cluj-Napoca, Cluj-Napoca, RomaniaTechnical University of Cluj-Napoca, Cluj-Napoca, RomaniaAs 3D depth cameras become more affordable, the additional depth information proves useful in industrial setups. However, with a multiview camera configuration, the complete coverage of a 3D model can only be accurately captured from several views. For this task, the next-best-view (NBV) estimation is essential to minimize the actions required for a complete reconstruction. Traditional NBV methods for 3D data are often sensitive to noise and lack rotation invariance. Learning-based methods can make estimations based on current camera information, significantly improving the accuracy of a task. These methods are generally limited to one step, as multi-step planning requires estimating sensor observations. A method that can predict and simulate multiple steps can significantly improve performance when a single step is insufficient. We show that a convex hull of the object can be used to predict n-steps of NBV in a coverage reconstruction setting. The convex hull is used as a proxy for the real geometry of the object in selecting views and requires fewer points sampled from the ground-truth geometry. We propose a volumetric information gain (VIG) score that measures the similarity between the convex hulls obtained from the partial point cloud from selected views and the ground truth point cloud. We retrain an NBV network to predict the views that lead to a point cloud with the highest VIG score, and demonstrate that spherical harmonic preprocessing can mitigate the effect of noise, thereby making the network faster. The convex hull is afterwards used to simulate views, and a set covering optimisation is applied to make a one-shot selection of the best views for surface coverage reconstruction. This method is robust to changes in view space, as shown in the tests applied on Shapenet models along with more complex synthetic shapes from HomebrewDB, Stanford 3D Scanning Repository, and Linemod.https://ieeexplore.ieee.org/document/11017667/Pose estimationspectral analysisplanning3D reconstructionpoint cloud compression |
| spellingShingle | Alexandru Pop Levente Tamas Spherical Harmonics for Robust Next-Best-View Estimation IEEE Access Pose estimation spectral analysis planning 3D reconstruction point cloud compression |
| title | Spherical Harmonics for Robust Next-Best-View Estimation |
| title_full | Spherical Harmonics for Robust Next-Best-View Estimation |
| title_fullStr | Spherical Harmonics for Robust Next-Best-View Estimation |
| title_full_unstemmed | Spherical Harmonics for Robust Next-Best-View Estimation |
| title_short | Spherical Harmonics for Robust Next-Best-View Estimation |
| title_sort | spherical harmonics for robust next best view estimation |
| topic | Pose estimation spectral analysis planning 3D reconstruction point cloud compression |
| url | https://ieeexplore.ieee.org/document/11017667/ |
| work_keys_str_mv | AT alexandrupop sphericalharmonicsforrobustnextbestviewestimation AT leventetamas sphericalharmonicsforrobustnextbestviewestimation |