Motor imagery brain electrical signal recognition method based on dual-tree complex wavelet energy difference

Abstract

The invention discloses a motor imagery brain electrical signal recognition method based on dual-tree complex wavelet energy difference. Energy difference features of brain electrical signals are calculated mainly through dual-tree complex wavelet decomposition and reconstruction, and are classified and judged through a sign function. The method includes the steps of brain electrical signal collection and preprocessing, dual-tree complex wavelet decomposition and detail coefficient extraction, dual-tree complex wavelet reconstruction, reconstruction component energy averaged value calculation and sign function classification and recognition. Results show that dual-tree complex wavelet conversion effectively overcomes the defects of poor aliasing resistance, translation sensitivity and the like of discrete wavelet transformation, and good classification and recognition results can be obtained through extracted energy averaged value difference features. Compared with a traditional classification algorithm, the feature classification algorithm based on features of the sign function is easier to design, low in complexity, high in calculation speed, suitable for the development direction of a Brain-Computer Interface (BCI) system and favorable for real-time application of the BCI system.

Description

technical field [0001] The invention belongs to the field of electroencephalogram signal processing, and relates to motor imagery EEG signal recognition in a brain-computer interface, in particular to a motor imagery EEG signal recognition method based on dual-tree complex wavelet energy difference. Background technique [0002] Stroke is currently one of the three major causes of death for humans, with approximately 20 million new patients worldwide every year, and 25% of patients become disabled after stroke. Neuromuscular stimulation systems and mechanically powered exoskeletons have been used to restore hand motor function in chronically paralyzed patients. The above-mentioned exercise assisting device is driven by a command signal, and the command signal needs to reflect the patient's exercise intention. If hand movement intentions could be detected directly from the brain, brain activity could be used to directly control motor assistive devices. At present, for the s...

AI Technical Summary

Problems solved by technology

[0006] In view of the poor anti-aliasing performance of the classical discrete wavelet transform in the prior art, it is impossible to accurately separate adjacent frequency bands, which ultimately leads to the defect that the extracted time-frequency features cannot correctly reflect the imaginary movement of the brain. A dual-tree with excellent anti-aliasing performance is proposed. The complex wavelet transform extracts the energy characteristics of the signal in the frequency band related to the motor rhythm of the electrode leads C3 and C4, and calculates the difference between the two, and finally classifies the motor intention EEG signal recognition method according to the sign function

Images