MEMD tensor linear Laplacian discrimination-based electromyographic feature extraction method

Abstract

The invention relates to an MEMD (Multivariate Empirical Mode Decomposition) tensor linear Laplacian discrimination-based electromyographic feature extraction method. According to conventional electromyographic feature extraction methods, a scatter matrix is often calculated by using a Euclidean distance based on vectors, so that the problems of between-class scatter matrix singularity, projectiondirection finiteness and the like exist. According to the MEMD tensor linear Laplacian discrimination-based electromyographic feature extraction method, multi-dimensional information of time-frequency-space domains and the like of a signal can be considered at the same time based on data representation of a tensor structure. Firstly a multi-channel electromyographic signal is filtered by using anMEMD method; secondly four-order tensor data with time, space, frequency and tasks is constructed by adopting wavelet packet transform; thirdly an optimal projection matrix is searched for by adopting a tensor linear Laplacian discrimination method to obtain a high-dimensional tensor feature with a relatively high discrimination degree; fourthly the high-dimensional tensor feature is subjected tomatrix processing and dimension reduction; and finally electromyographic features are identified by adopting a conventional classification method. The method has a wide application prospect in the field of man-machine interaction of rehabilitation robots and the like.

Description

technical field [0001] The invention belongs to the field of electromyographic signal processing, and relates to a method for extracting features of electromyographic signals, in particular to a method for extracting features for human-computer interaction. Background technique [0002] The development of robot technology is strongly promoting the application of robots from industrial production to military, medical, service and other fields. In the future society, the communication between humans and robots and even the direct combination of each other's bodies will become more and more frequent, as the information channel connecting humans and robots Advanced human-computer interaction (HRI) technology will surely play a vital role in human life. The traditional human-computer interaction method based on program control shackles the robot's autonomous adaptability, and is difficult to apply to robot systems that need to be directly integrated with the human body, such as b...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims at the traditional EMG feature extraction methods are often based on vectors, and use the Euclidean distance to calculate the dispersion matrix, so there are problems such as the singularity of the dispersion matrix between classes and the limited projection direction

Images