WIMI Hologram Academy: EOG-based Human-Computer Interaction Technology in Virtual Reality
WIMI Hologram Academy, working in partnership with the Holographic Science Innovation Center, has written a new technical article describing their exploration of EOG-based human-computer interaction technology in Virtual Reality.
Virtual reality technology, as an advanced computer simulation technology, has been widely used in many fields. In the medical and rehabilitation fields, for disabled patients with physical injuries, virtual reality-based rehabilitation training can provide them with interesting and comprehensive training, accurate sensory feedback, and a safe training environment. According to the patient’s psychological state and condition, virtual reality technology can select corresponding rehabilitation training scenarios and tasks, stimulate and maintain the patient’s initiative in rehabilitation training with various forms of feedback, and improve the rehabilitation effect.
Immersion, interactivity and conceptualization are the three basic features of virtual reality systems. Traditional virtual reality interaction methods mainly include VR handles, data gloves, motion capture, etc. With the continuous development of information technology, some new methods of human-computer interaction technology for virtual reality have emerged. For example, interaction based on neuron-electric signals such as EEG and EMG. These emerging human-computer interaction technologies have greatly enhanced the immersive experience of virtual reality. Scientists from WIMI Hologram Academy of WIMI Hologram Cloud discussed in detail a new type of virtual reality interaction technology, the electro-oculography (EOG)-based interaction.
1. EOG-based virtual reality interaction technology
The EOG signal-based virtual reality interaction system mainly includes three parts, which are the signal acquisition part, the EOG signal processing part and the virtual reality scene part. When the system works, the user receives the excitation signal from the virtual reality scene in real time, and the user makes corresponding eye movements according to the excitation signal. The signal acquisition equipment collects the user’s EOG signal in real time, and then converts it into control commands for the virtual reality scene through a series of signal processing, so that the virtual reality scene executes the corresponding commands and gives feedback to the user in a visual way.
1.1 Electro-ocular signal acquisition
The EEG signal is caused by the difference in electrical potential between the cornea and the retina, and can be used to reflect eye movements, with amplitudes generally ranging from 0.4 to 10 mV. Compared to EEG signals, the acquisition of EEG signals is relatively simple and convenient, usually requiring only a few electrodes. Generally speaking, six electrodes (e.g., A.B.C.D.E.F) can be used to acquire EEG signals. Electrodes A and D are located above and below the eye and are used to collect the vertical EEG signals, which are mainly generated by up and down movements of the eye or blinking, while electrodes B and C are located on the left and right side of the eye and are used to collect the horizontal EEG signals, which are mainly generated by left and right movements of the eye. The EOG signal usually has higher amplitude and more stable waveform shape compared with EEG signal, so it is easier to detect.
1.2 Electro-ocular signal processing
The EEG signal processing mainly includes several major steps, such as signal pre-processing, feature extraction, waveform detection and classification recognition.
(1) Signal pre-processing
There are many methods to pre-process the original EEG signal, including signal amplification, baseline calibration, artifact removal, down-sampling and other methods. The frequency band in which the EEG signal is located is low, and the original EEG signal is generally mixed with other bio-electric signals and external power frequency noise interference signals, so in the pre-processing link, generally through low-pass filtering, wavelet transform and other methods to attenuate or eliminate the baseline drift and high-frequency noise brought about by the interference.
Recommended AI News: Getty Images Revolutionizes Visual Communications with VisualGPS Insights Launch
(2) Waveform detection
The basic principle of waveform detection is to determine whether the subject performed a valid single blink by the amplitude and signal duration of the feature vector F (i.e., the differential EEG waveform) obtained after the above pre-processing. Experiments indicated that there was a very distinct peak-valley characteristic in the differential EEG signal waveform, and the valley appeared after the peak.
To detect the extracted feature vector F after each excitation signal, we first need to find the locations of the peaks and valleys in F (the largest value in the extreme value of the feature vector F is considered as the peak, and the time point corresponding to the peak is set as t-peak; the smallest value in the extreme value is considered as the valley, and the time point corresponding to the valley is set as t-valley). Then, for the obtained feature vectors F corresponding to different excitation signals, the interval time t-interval between the wave peak and the wave trough and the accumulated energy e in each feature vector are calculated separately, and finally, the existence of blinking action is judged by Eq.
(3) Feature extraction
Before feature extraction, a single EEG data segment is generally extracted, and the length of the data segment is designed according to the actual situation, and the feature vector is also generally extracted for a single cycle of the EEG signal. It should be noted that the extracted feature vectors should effectively represent the characteristics of the EEG signal, have good differentiation and independence, and be easy to compute. The methods of EEG signal feature extraction include shape feature extraction method based on signal waveform, wavelet transform method, etc.
(4) Classification recognition
Currently, the most common method used to classify the features of EEG signals is the threshold method. In addition, methods such as support vector machines, BP neural networks, and linear discriminant analysis can also be applied to the classification of EEG signals. Each method has its own advantages and limitations, and the most suitable processing method should be chosen according to the actual situation.
[To share your insights with us, please write to email@example.com]