Expanded section Gaussian-mixture filter

Abstract

The invention relates to an expanded section Gaussian-mixture filter, which comprises a Gaussian approximation module, a prediction module, a Gaussian-mixture reduction module, a section Gaussian-mixture approximation module and a filter module, wherein the Gaussian approximation module is used for performing Gaussian-mixture approximation on a probability density function of a received non-Gaussian noise; the prediction module is used for predicting a state of an input signal containing the Gaussian-mixture noise, and the predicted state probability density function is in a Gaussian-mixture form; the Gaussian-mixture reduction module is used for reducing the predicted state probability density function; the section Gaussian-mixture approximation module is used for approximating the predicted state probability density function to a section Gaussian-mixture form; and the filter module is used for updating the state of the input signal. Due to the adoption of the expanded section Gaussian-mixture filter, the state estimation problem of a mixed linear/nonlinear system in a non-Gaussian noise environment can be solved, and a state device in the non-Gaussian noise environment is realized by a paralleling section Gaussian-mixture filter.

Description

technical field [0001] The invention relates to a digital filter, in particular to a sliced ​​Gaussian mixture filter. Background technique [0002] A digital filter is an algorithm or device that changes the signal spectrum and completes the filtering function through the operation and processing of digital signals. A digital filter is a discrete-time system that can be implemented with computer software. Digital filters have the advantages of high precision, high reliability, programmable change of characteristics or multiplexing, and easy integration. Digital filters are widely used in speech signal processing, image signal processing, medical biological signal processing and other application fields. [0003] In a noisy environment, it is usually necessary to denoise the system, and the commonly used denoising method is a filtering method. The Kalman filter has optimal filtering performance for linear Gaussian systems, but the performance of the Kalman filter is signi...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the state estimation problem of the mixed linear / nonlinear system in the non-Gaussian noise environment, realize the state machine in the non-Gaussian noise environment with a parallel sliced ​​Gaussian mixture filter, and realize an extended sliced ​​Gaussian hybrid filter

Images