A Survey of Data Fitting Based on Moving Least Squares
Download as PDF
Lin Shaofu, Xia Sibin
Moving least squares(MLS) is a common method of data fitting, and it has a high degree of flexibility and precision that is significantly better than other fitting methods. This paper introduces the principle of MLS, enumerates the important research progress in recent years at home and abroad, and analyzes the advantages of the method applied in data fitting tasks. MLS also has problems such as susceptibility equations, support domain and weight function selection relying on empirical judgment. The researchers put forward some strategies for the problem, but they have not solved it fundamentally. For the future research direction, this paper suggests that the researchers should study in depth from the mathematical theory, convergence, error analysis and performance comparison of MLS method and its improved methods.
Data Fitting, Moving Least Squares, Meshless Method