Abstract:We consider the problem of reconstructing 3D objects via meshfree interpolation methods. In this framework, we usually deal with large data sets and thus we develop an efficient local scheme via the well-known Partition of Unity (PU) method. The main contribution in this paper consists in constructing the local interpolants for the implicit interpolation by means of Rational Radial Basis Functions (RRBFs). Numerical evidence confirms that the proposed method is particularly performing when 3D objects, or more in general implicit functions defined by scattered data, need to be approximated.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to efficiently and accurately reconstruct 3D objects through the mesh - free interpolation method. Specifically, the author focuses on how to approximate 3D objects using reasonable methods when dealing with large - scale data sets. To this end, they developed a local partition of unity (PU) method based on rational radial basis functions (RRBFs).
### Problem Background
In computer - aided design and computer graphics, reconstructing 3D objects is a common problem. Usually, such problems involve approximating the surface defined by point - cloud data (i.e., an unordered set of points in 3D space). Since these data sets are usually very large, researchers tend to use mesh - free methods to handle these problems. In particular, when dealing with large amounts of data, traditional global interpolation methods may lead to difficulties in solving large - scale linear systems, so a more efficient method is required.
### Main Contributions
The main contributions of this paper are as follows:
1. **Introducing RRBFs**: To improve the stability and accuracy of local interpolation, the author proposes using rational radial basis functions (RRBFs). RRBFs can better handle functions with steep gradients and are more numerically stable.
2. **Combining with the PU method**: By decomposing the original problem into multiple small problems, the partition of unity (PU) method is used to reduce the computational complexity. The PU method can divide the original reconstruction domain into multiple sub - domains, so that only medium - scale linear systems need to be solved.
3. **Optimizing the algorithm**: For each sub - domain, the RRBF expansion ultimately boils down to a maximum eigenvalue problem, which can be efficiently solved by the so - called deflation - accelerated conjugate gradient (DACG) algorithm.
### Method Overview
The specific steps of this method are as follows:
1. **Generating additional off - surface points**: To achieve implicit interpolation, additional off - surface points are added to construct a three - dimensional interpolation problem.
2. **Constructing the PU structure**: The PU structure is constructed according to the divided sub - domains.
3. **Calculating the local RRBF approximation**: Calculate the local RRBF approximation for each sub - domain.
### Numerical Experiments
The author verified the effectiveness of the proposed method through a series of numerical experiments. The experimental results show that the RRBF - PU method performs well when dealing with both known implicit functions and unknown 3D objects, and especially when dealing with large - scale data sets, its performance is better than that of the traditional classical method based on local RBF interpolation (such as RBF - PU).
### Conclusion
In conclusion, this research shows that the RRBF - PU method is an efficient and robust tool for approximating 3D objects. Future work will further study how to combine this method with existing stable methods and further explore issues such as the Lebesgue constant.
Hopefully, this summary can help you understand the core content and main contributions of the paper. If you have any further questions or need more detailed information, please feel free to let us know!