1. The introduction

The human-computer interaction of virtual reality refers to the information transmission between users and virtual world objects generated by computers in a portable and natural way. Establishing a more natural and harmonious man-machine environment through two-way perception between users and virtual environment is the core link of virtual reality to provide users with experience and move towards application. With the continuous development of information technology, the human-computer interaction mode of virtual reality has gradually expanded from traditional input devices such as handle and keyboard to interactive modes such as voice, posture and physiological signals. This paper introduces a new virtual reality interaction technology, motor imagination brain-computer interface technology based on eeg.

2. Motor imagination brain-computer interface interaction technology based on EEG

Human computer interaction technology based on EEG is a process of obtaining eeg signals from human scalp by signal acquisition equipment, and then transforming different EEG characteristics into corresponding control commands to realize human computer interaction. This technology is called brain-computer Interface (BCI), and it directly uses human Brain electrical information to interact with virtual objects in virtual scenes, without the involvement of the user's hands, feet or any other muscle movements.

According to the different generating principles of eeg signal, brain-computer interface technology can be roughly divided into five categories: Brain-computer interface based on Motor Imagery (MI), steady-state Visual Evoked Potential, Cortical Cortical Potential (SCP) based BRAIN computer interface (SSVEP) and Motion-onset Visual Evoked Potential (MOTION-onset Visual Evoked Potential) MVEP).

2.1 Principle of brain-computer interface technology based on motor intention imagination

Motor Imagery potential (MI) refers to the specific brain rhythm generated when the subject imagines exercising, which is represented by the eeg potential oscillating at a specific frequency, including mu band (8-13 Hz), Beta band (13-30 Hz), etc. When subjects imagine a limb moving, the energy in these bands decreases, a phenomenon known as Event Related Desynchronization (ERD), and conversely, when subjects stop imagining movement, the energy in these bands gradually rises again. The corresponding phenomenon is called Event Related Synchronization (ERS). This change in frequency band energy usually occurs in the contralateral brain region of the imagined limb, so it is possible to identify left or right limb movement by frequency analysis of the subjects' eeg signals. As shown in Figure 1, when subjects imagine left hand movement, ERD signals in the right brain region are more obvious. Therefore, when calculating the power spectral density of eeg signals, it is found that the energy in the left brain region is significantly higher than that in the right brain region, and vice versa. At present, Common Spatial Pattern (CSP) algorithm is commonly used for feature extraction of motion imagery signals. CSP algorithm is a supervised method, that is, the training data set is marked and the class of each data vector is known. For example, brain signals were collected while subjects performed two different tasks, such as left - and right-handed motor imagery. CSP searches for spatial filters to maximize the variance of the filtered data with one class and minimize the variance with the other class. Therefore, the obtained feature vectors enhance the differences between the two classes. The variance of the eeg signal obtained by filtering a particular frequency band corresponds to the energy of that frequency band, so CSP essentially maximizes the distinguishing degree of the features used by BCI.

The generation of P300 and SSVEP signals mentioned above requires external stimulation to induce, which belongs to induced eeg. In order to realize the stable induction of eeg signal, it often needs a sustained time (generally a few seconds or so). In contrast, the generation of motor imagination signal does not need any external stimulation, which belongs to spontaneous EEG. Users can output continuous control commands through continuous motor imagination, and transmit the command to virtual reality environment, so as to realize virtual human-computer interaction based on brain computer interface.

Figure 1. ERD in right-handed imagination (left: power spectrum of C3 electrode in left-handed imagination, middle: ERD topological distribution, right: power spectrum of C4 electrode in right-handed imagination)

2.2 Motion image-virtual reality system structure

Motor imagination Brain Computer Interface (MI-BCI) system is an interactive communication system combining hardware and software between the brain and the controlled object. The basic MI-BCI system consists of the following parts: Brain electrical signal acquisition, data processing and analysis, virtual scene stimulus control and feedback, etc., among them, the eeg acquisition equipment acquisition subjects for specific tasks of thinking brain electrical signal, then USES the eeg signals processing analysis algorithm analytic movement intentions, directly to the subjects of thinking into the virtual reality scene object control instruction, The feedback link shows the current state of the subjects and the implementation of the system in real time. According to the feedback information, the subjects actively strengthen or weaken the intensity of motor imagination, adjust the generation of spontaneous EEG signals, and improve the quality and control accuracy of EEG signals. See Figure 2.

FIG. 2 Principle structure of human-computer interaction system based on motion imagination

(1) Eeg signal acquisition

Eeg signal acquisition provides the input part for MI-BCI system. Research on the brain mechanism of motor imagination based on EEG confirms that most of the brain regions activated by motor imagination and motor execution overlap. The unit of eeg signal is uV. It is collected by placing a certain number of electrodes in a specific area of the brain to collect the eeg signal of the brain during the movement of imagination. The signal is converted into digital signals that can be recognized and processed by the computer through collectors, amplifiers, filters and analog-digital converters.

In the motor imagination experiment, the subjects' brains perform different thinking tasks, and the spontaneous eeg signals produced by different rhythms show different activity states. According to frequency range, eeg signals can be divided into four common rhythms, as shown in Table 1. The eeg signal analysis of motor imagination mainly studied the signals of MU and Beta rhythm, and classified the EEG signals of different thinking tasks according to the energy difference of MU /Beta rhythm during left and right hand motor imagination.

Table 1. Eeg rhythm

(2) Eeg signal processing and analysis

Eeg signal has the characteristics of randomness, nonlinearity and non-stationarity, and contains noise interference of different frequencies. Before the analysis of eeg signal, signal preprocessing is needed to screen out effective rhythm change data of motor imagination. By comparing the time-frequency characteristics of motor imagination eeg signals, feature extraction algorithm was used to extract feature vectors that can reflect different thinking tasks to the maximum extent, and the best classifier algorithm was used to identify different thinking states and convert them into control instructions. According to the results of offline analysis, the algorithm combinations suitable for different subjects and the parameter information of each algorithm are determined and saved separately, and the corresponding information can be directly loaded in the online experiment.

(3) Virtual reality control and feedback

The purpose of the study of motor imagination BRAIN-computer interface system is to realize the direct communication between human brain and controlled equipment. The feedback link of the system is used to prompt subjects to adjust their own state, improve the quality of motor imagination EEG signals, and finally obtain higher classification and recognition rate. Through the design of VR system, the recognition results of MI-BCI system are received in real time during the online control, and the intention of the subjects is converted into the control instructions of VR system through corresponding instruction rules. The implementation effect of VR scene is fed back to the subjects in real time, and the subjects adjust their motion imagination thinking state in real time through the interaction of VR. To get the best control effect.

3. The conclusion

The brain-computer interface based on EEG can interact with virtual reality scene objects according to people's independent thinking activities, without the participation of hands, feet and any other muscle movements, which greatly improves the user's immersive experience. With the continuous development of brain computer interface technology, virtual reality human-computer interaction technology based on brain computer interface will gradually mature, and will play an important value in the fields of medical treatment, education, entertainment and so on in the future.