Design And Implementation Of BCI System Based On 3D Visual Stimulator

BCI (Brain-Computer Interface, BCI) is a new and interdisciplinary human-machine interface technology developed recently. It is a communication system which is independent of people's normal peripheral nerve and muscle tissue. BCI research has a clear background in the clinical application. In the field of medical rehabilitation, it provides a new method of communication with the outside world for these muscle impaired patients. Therefore, the research and development of BCI systems is of great significance.After the consultation of a large number of domestic and international researches, the principle of a BCI system based on 3D visual stimulator has been proposed in this dissertation. The work for this paper is to design and implementation visual stimulator, then mu rhythm Event-related desynchronization (ERD) corresponding to brain activities from sensory-motor cortex in the process of the left and right hand motor imagery has been investigated, specific contents are as follows:①Design and implement 3d visual stimulator. This paper has adopted a process control strategy based on the key objective action, and it has ensured system's real-time, availability and scalability by controlling the static model database to generate motion. In order to realize 3D model rendering and flexible control, OpenGL has been identified as the core technology to develop system. In the process of implementing system, to reduce the memory space utilization of the system by separately storing model vertices and faces; to enhance the rendering effect of model surface by applying the algorithm of Gouraud shading; to solve the problem of real-time display by designing a multi-buffer mechanism with a model cache buffer; to fulfill the high-precision control of action by utilizing multimedia timer. Results have been demonstrated that the visual stimulator is satisfied with the request of practice application.②Extract feature of Motor imagery EEG. This paper focuses on the ERD of mu rhythm during the process of motor imagery. By comparing mainstream methods of time-frequency analysis, this paper has adopted Hilbert Huang Transform (HHT) to obtain the marginal spectrum and time domain amplitude spectrum to make the time-frequency analysis of EEG. Furthermore, this paper has investigated selective preference of frequency-band and time-window to achieve individuation analysis on the ERD of mu rhythm according to Fisher criterion. To further enhance noise ratio of the EEG signal, an improved CSP algorithm has been applied. Under the condition for the small number of lead, after the CSP filter, the scalar feature of EEG displays instability, the feature vector space of EEG is too large if directly combine the filter signal. To solve the two problems, this paper has made a further improvement in feature selection, which combined with time-domain energy analysis to extract reduced dimension feature by sliding window③Make experiments on the BCI system. This paper has designed three different ways of visual stimulus: no goal-oriented grip, goal-oriented grip on static object, and goal-oriented grip on dynamic object. In the end, this paper has analyzed results of timing accuracy and feature classification.To sum up, this paper has investigated on brain computer interface system based on 3d visual stimulator, and laid a foundation for the cross from offline BCI system to online BCI system.