Study On Multiple Mobile Target Selection Based On Brain-Computer Interface

Brain computer interface(BCI)is an emerging technology that allows communication with the outside directly instead of through the peripheral nerves and muscles.It has been gradually applied to various kinds of information interaction and physical equipment control systems.In the practical application system,user needs to select targets from the environment.However the traditional BCI paradigms are generally static,users need to switch between the BCI paradigm and the real scene,and when facing the mobile target selection,the static BCI paradigm is incoherent,the flexibility is prone to be constrained,and the information provided are also limited.In order to apply BCI technology to mobile target selection,this paper attempts to embed two kinds of evoked BCI into dynamic scene,to realize the coherent selection of multiple mobile targets.We set the pedestrians in the video as the targets to be selected,and designed mobile target selection paradigms based on the SSVEP tried and P300 respectively,the main contents of the thesis are as follows:A BCI moving target selection system is constructed,we use the induced BCI method to select the moving targets,and design multiple mobile target selection paradigm based on SSVEP and P300 respectively.The pedestrian of the video is selected as the moving target,and the target detection algorithm is object detection with discriminatively trained part based models,and the tracking algorithm is a global multi-object tracking algorithm based on generalized minimal group problem.the visual stimuli of the induced BCI are superimposed on the perdestrian targets to induce the corresponding neural signal components to achieve the target selection.In the exploring on multiple mobile target selection based on SSVEP BCI approach,for the SSVEP stimulus design,stimulus patterns were determined from the stimulus shape,size,frequency and flicker time.Due to the difference of the subjects,the optimum single target selection time was determined for each participant by experiments,with the accuracy and information transfer rate as the assessment criterion.In th exploring on multiple mobile target selection based on P300 BCI approach,the color change number overlapped on the pedestrian target is taken as the stimulus event.The classifier and the optimal EEG signal acquisition channels were trained for each subject through the offline experiments.The experimental results show that the accuracy and response time of both two approaches can meet the requirements of system availability for practical application,which validates the feasibility of BCI selection for moving targets.Compare the two methods,the SSVEP method has a fast response and strong maneuverability,but the adaptability of the method is relatively large.By contrast,the P300 method has slow speed,low rate of information transmission,but the anti-jamming performance is stronger and adaptability is better.Compared with the traditional static BCI,the BCI dynamic paradigm implemented in this paper reduces the limitation of the user interaction interface,so that the user can grasp the global information while paying attention to the environmental target,which make the system more flexible and versatile,and can be extended to other dynamic environment applications when combined with different image processing algorithms.Thus the coherent multiple mobile target selection based on BCI method has broad application prospects.