A Real-Time Multi-Model Parametric Representation of Point Clouds
Yuan Gao, Wei Dong
AI summary
Problem
High-accuracy parametric models are too computationally heavy for real-time robotic use, while fast real-time methods lack the geometric flexibility to accurately capture complex environments.
Approach
The method segments point clouds using an integrated hierarchical Gaussian mixture model, then merges flat clusters into planes or curved B-spline surfaces. A 2D voxel-based technique efficiently describes boundaries for all primitives, while unstructured regions are retained as Gaussian distributions.
Key results
- 3.78× faster curved surface detection compared to state-of-the-art methods
- Twofold accuracy improvement over pure Gaussian mixture models
- Real-time processing at 36.4 fps on low-power computers
- Unified representation handling unstructured, planar, and curved regions simultaneously
Why it matters
Enables efficient, high-fidelity environmental modeling for memory-constrained and bandwidth-limited robotic systems.
Abstract
In recent years, parametric representations of point clouds have been widely applied in tasks such as memory- efficient mapping and multi-robot collaboration. Highly adap- tive models, like spline surfaces or quadrics, are computa- tionally expensive in detection or fitting. In contrast, real- time methods, such as Gaussian mixture models or planes, have low degrees of freedom, making high accuracy with few primitives difficult. To tackle this problem, a multi-model parametric representation with real-time surface detection and fitting is proposed. Specifically, the Gaussian mixture model is first employed to segment the point cloud into multiple clusters. Then, flat clusters are selected and merged into planes or curved surfaces. Planes can be easily fitted and delimited by a 2D voxel-based boundary description method. Surfaces with curvature are fitted by B-spline surfaces and the same boundary description method is employed. Through evaluations on multiple public datasets, the proposed surface detection exhibits greater robustness than the state-of-the-art approach, with 3.78 times improvement in efficiency. Meanwhile, this representation achieves a 2-fold gain in accuracy over Gaussian mixture models, operating at 36.4 fps on a low-power computer.