Block selection common space mode feature extraction method for motor imagery electroencephalogram

Abstract

The invention discloses a block selection common spatial pattern feature extraction method for motor imagery electroencephalogram, which comprises the following steps of: firstly, preprocessing original data in a mode of constructing a data block by dividing a frequency band by each channel, and secondly, performing correlation calculation on each block to obtain an index Fisher ratio for representing classification performance; then, selecting data blocks according to the indexes and a reasonable threshold value; and finally, carrrying out feature extraction and classification on data formedby the optimal blocks by using CSP and SVM. The selected blocks can effectively avoid difference between different individuals and different channels, the channels to which the blocks belong are reasonable in distribution and moderate in number, the classification performance of a BCI system is improved to a certain extent, and a new thought is provided for feature extraction of electroencephalogram signals.

Description

technical field

[0001] The invention belongs to the field of pattern recognition, and proposes a block selection-co-space model feature extraction method (Block-selection for CSP, BS-CSP), which is used for task classification of a BCI system based on motor imagery. By dividing frequency bands for each channel to construct data blocks, and using the Fisher ratio calculated by the time-frequency characteristics of each block to select the optimal block, the simultaneous selection of channels and frequency bands is realized, and the redundancy between channels and frequency bands is reduced. Then use the common space pattern (Common Space Pattern, CSP) and support vector machine (Support vector machine, SVM) to perform feature extraction and classification on the data composed of the optimal block. Background technique

[0002] As a bridge between humans and computers, the brain-computer interface (Brain-computer interface, BCI) can generate control commands through the recogn...

Examples

Embodiment Construction

[0041] The method for feature extraction of block selection and co-space pattern based on motor imagery EEG of the present invention will be described in detail below in conjunction with the accompanying drawings. Such as figure 1 , the implementation of the present invention mainly includes 4 steps: (1) multi-channel EEG signal acquisition and preprocessing; (2) correlation calculation of blocks; (3) block selection; (4) CSP feature extraction and SVM classification. Assuming that the EEG data collected in the experiment has K channels, the data obtained by dividing the EEG of each channel into frequency bands is called a block. Each channel is divided into S frequency bands of equal bandwidth, and then K×S blocks are generated. Record the data of the block corresponding to the sth frequency band of the kth channel as x (ks) (n), k=1, 2,..., K, s=1, 2,..., S, n=1, 2,..., N, where N is the number of sampling points for each block. Assuming that the subject conducts a total ...

technical field

[0001] The invention belongs to the field of pattern recognition, and proposes a block selection-co-space model feature extraction method (Block-selection for CSP, BS-CSP), which is used for task classification of a BCI system based on motor imagery. By dividing frequency bands for each channel to construct data blocks, and using the Fisher ratio calculated by the time-frequency characteristics of each block to select the optimal block, the simultaneous selection of channels and frequency bands is realized, and the redundancy between channels and frequency bands is reduced. Then use the common space pattern (Common Space Pattern, CSP) and support vector machine (Support vector machine, SVM) to perform feature extraction and classification on the data composed of the optimal block. Background technique

[0002] As a bridge between humans and computers, the brain-computer interface (Brain-computer interface, BCI) can generate control commands through the recogn...

Examples

Embodiment Construction

[0041] The method for feature extraction of block selection and co-space pattern based on motor imagery EEG of the present invention will be described in detail below in conjunction with the accompanying drawings. Such as figure 1 , the implementation of the present invention mainly includes 4 steps: (1) multi-channel EEG signal acquisition and preprocessing; (2) correlation calculation of blocks; (3) block selection; (4) CSP feature extraction and SVM classification. Assuming that the EEG data collected in the experiment has K channels, the data obtained by dividing the EEG of each channel into frequency bands is called a block. Each channel is divided into S frequency bands of equal bandwidth, and then K×S blocks are generated. Record the data of the block corresponding to the sth frequency band of the kth channel as x (ks) (n), k=1, 2,..., K, s=1, 2,..., S, n=1, 2,..., N, where N is the number of sampling points for each block. Assuming that the subject conducts a total ...

Images