Steady-state visually induced EEG identification method and system based on deep learning

Abstract

The invention belongs to the technical field of identity information identification, and discloses a steady-state visually induced EEG identity identification method and system based on deep learning; acquisition of EEG signals; denoising: processing the collected EEG signals through EMD to obtain denoising EEG signal after noise; the extracted data is divided into three samples, and fast Fourier transform is performed on each sample data to obtain EEG data in the frequency domain; band-pass filtering; the filtered signal is sampled and processed at a frequency of 2048Hz; Build a deep network for training; identification: identify the purpose of the subject. The present invention is suitable for brain-computer interface devices with steady-state visual induction. As the number of people entered in the system increases, the training data increases accordingly. Based on the characteristics of the deep network, the recognition accuracy of the system tends to be higher. of stability.

Description

technical field [0001] The invention belongs to the technical field of identity information identification, and in particular relates to a steady-state visually induced EEG identity identification method and system based on deep learning. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: the research and development of identity recognition systems are very important for social life and personal daily life. Technology, that is, through the close combination of computer and high-tech means such as optics, acoustics, biosensors and biostatistics principles, using the inherent physiological characteristics of the human body (such as hand shape, fingerprints, facial features, iris, retina, etc.) and behavioral characteristics (such as handwriting , voice, gait), etc. to carry out personal identity identification. EEG signals are generated by the transmission of information in the form of ions by neurons in the brain...

AI Technical Summary

Problems solved by technology

[0004] (1) The EEG signal has the characteristics of low signal-to-noise ratio, randomness, and features that change with the surrounding environment, and it is difficult to extract features from the EEG signal

Therefore, the features extracted at a point in time may change and become unusable after a period of time

[0005] (2) The existing technology cannot efficiently extract the electrode spatial distribution information and time-frequency information in the EEG signal

Therefore, it is impossible to identify EEG signals in real time, and most of the literature is only an identification test for a single offline EEG data.

[0006] (3) Deep learning can extract features different from existing methods, and existing technologies cannot maturely apply deep learning to the field of EEG

Therefore, the stability of the manually extracted features over time cannot be guaranteed, and the accuracy of the identification system based on the SSVEP signal cannot be guaranteed.

[0007] The difficulty and significance of solving the above technical problems: Difficulty: firstly, the signal-to-noise ratio of the EEG signal is low, so it is necessary to fully extract the information contained in the EEG signal; secondly, if the electrode information of the EEG signal and the The combination of time-frequency information and extraction will greatly increase the accuracy of EEG identification; finally, the application of deep learning in the image field has become increasingly mature, but how to use deep learning to better extract EEG signals cannot be extracted by traditional methods. The characteristics of the identity recognition system increase the recognition rate

Images