Electroencephalogram classification method based on multi-classifier integration

Abstract

The invention discloses an electroencephalogram signal classification based on multi-classifier integration. The method includes the following steps: performing electroencephalogram signal acquisition and preprocessing; performing feature extraction on electroencephalogram signals through utilization of a time domain analysis method, an autoregressive model and a discrete wavelet transform method in a combined manner; establishing an individual support vector machine classifier model, and converting outputs of the support vector machine classifiers into probability outputs; performing multi-classifier integrated electroencephalogram mode classification, and using a D-S evidence theory to fuse classified information of the three individual support vector machine classifier to obtain a final classification result. An experimental result shows that the method can improve the accuracy rate of motor imagery electroencephalogram signal classification.

Description

technical field [0001] The invention relates to the field of EEG signal processing and pattern recognition, in particular to the classification of motor imagery EEG signal patterns in brain-computer interfaces, and in particular to an EEG classification method based on multi-classifier integration. Background technique [0002] The brain-computer interface is a communication system for information transmission between the brain and external devices. It can convert the potential activity of the user's brain into control commands for external devices, thereby replacing body and language to communicate with the outside world. At present, motor imagery EEG collected from the human scalp is often used for non-invasive brain-computer interface control. The brain-computer interface based on motor imagery EEG mainly recognizes the user's motion intention through the analysis and processing of motor imagery EEG, and then converts the recognition results into control commands for exte...

AI Technical Summary

Problems solved by technology

[0004] At present, a single classifier is mostly used in the pattern recognition of the brain-computer interface, which makes it difficult to improve the classification accuracy, and the general performance is also poor. Through the ensemble learning method, constructing and combining multiple classifiers, it is possible to obtain better results than a single classifier. High accuracy and pan-China performance

Images