KEYWORDS: Point clouds, Data modeling, Education and training, Feature extraction, Singular value decomposition, Matrices, Error analysis, Evolutionary algorithms, 3D modeling, Semantics
The core problem of point cloud registration can be explained as computing rigid transformations to align two point clouds, which is a key technology used in popular fields such as three-dimensional reconstruction, robotics, and autonomous driving. We come up with a hybrid feature-based model inspired by Deep Closest Point (DCP) and Robust Point Matching using Learned Features (RPMNET). The main innovation of this model is to combine abstract features extracted by Dynamic Graph CNN (DGCNN) with Point Pair Features (PPF) as hybrid features for point cloud registration, after that, soft matching is performed between two point clouds, and then singular value decomposition (SVD) is applied to compute the rigid transformations. Besides, we adopt the ModelNet40 dataset for training and compare the trained model with DCPV2, Iterative Closest Point (ICP) and some other ICP variants, the comparison of results indicates our model performs better than the above methods in predicting the angle of rotation when rigid transformations occur. We also test our model on clean and noise-added test sets respectively to verify the robustness of it.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.