Motor imagery classification method based on optimal narrow band feature fusion

Abstract

The invention discloses a motor imagery classification method based on optimal narrow band feature fusion. According to the method, four-classification motor imagery tasks are integrated into four two-classification motor imagery tasks, then one optimal narrow frequency band is obtained for each two classes of motor imagery tasks, and four optimal narrow frequency bands are obtained in total; carrying out band-pass filtering on every two types of motor imagery electroencephalogram signals by utilizing an optimal narrow band, then carrying out feature extraction on the filtered electroencephalogram signals, and generating a result matrix with the dimension of 32 * 7; and constructing a deep convolutional neural network model, inputting a 32 * 7 result matrix, and outputting an electroencephalogram signal prediction category. According to the method, an optimal narrow band is automatically determined through a novel quadtree search tree, dynamic energy features are extracted through a common spatial pattern algorithm, finally, feature fusion is conducted on the multiple narrow bands through a convolutional neural network, and classification of multi-class motor imagery electroencephalogram signals is achieved.

Description

technical field [0001] The invention belongs to the motor imagery paradigm in the field of brain-computer interface, and relates to a motor imagery classification method based on optimal narrow-band feature fusion. Background technique [0002] Brain-Computer Interface (BCI) is a technology that establishes external information exchange and control between the human brain and computers or other electronic devices that does not rely on conventional brain information output pathways. Since the birth of BCI, the mainstream of its research and development has mostly been to control external devices and replace some missing functions of patients, for example, to realize the compensation of missing functions by controlling devices such as wheelchairs and spellers. As a non-invasive signal acquisition method, EEG is the main application method of BCI due to its safety and low cost, and motor imagery is the most widely studied paradigm of EEG. Signal classification is also the focu...

AI Technical Summary

Problems solved by technology

Two different input modes have their advantages and disadvantages. The end-to-end neural network can automatically learn useful features from the original data, but for small training data sets, it is difficult to obtain satisfactory results, and for data under different tasks set, the model needs to be adjusted according to relevant background knowledge

The feature input network is suitable for small data sets and is superior to traditional methods, but after feature extraction and then input to the network, some potential information will be lost, thus affecting classification performance

Images