Curve fitting method based on genetic algorithm and Gaussian mixture level set model

The invention provides a curve fitting method based on a genetic algorithm and a Gaussian mixture level set model. The method is applied to curve fitting of closed discrete data points and solves theproblems that fitting precision of the closed discrete data points is low, fitting processes of discrete data points of unknown data characteristics are complicated, and over-fitting is easily generated in a current curve fitting method. The particular scheme includes that the method includes the steps: firstly, building a two-dimensional Gaussian mixture model, and calculating Gaussian mixture model parameters according to the given closed discrete data points; secondly, collecting level set algorithm thoughts by the aid of the two-dimensional Gaussian mixture model, and building a closed curve to serve as an initial curve by intercepting data points of the same level set in the two-dimensional Gaussian mixture model; thirdly, representing deviation among fitting points and sample pointsby the aid of a least squa