The Design And Implementation Of Asynchronous HCI Algorithm Based On EOG

With the rapid development of Internet and wearable devices, the biological signal detection of human body and its application have become an active field currently. In which, the EOG-based human-computer interaction (HCI) and human activity recognition (HAR) are two new developing directions. The related researches involve many aspects, such as the detection and feature extraction of EOG, the pattern recognition and classification of eye movement, the realization of online system and other key technologies. However, in the real environment, due to the complexity of background noise and interference, the efficient feature extraction of the EOG waveform is difficult. Especially for the online implementation of EOG detection and analysis system, the core DSP algorithm should meet higher requirements in stability and real-time performance. Therefore, the research of EOG analysis and processing algorithm for eye movement recognition is of great theoretical and practical significance.In this thesis, the novel approaches of EOG signal processing and feature extraction are studied in the context of the EOG-HCI applications; The main work is as follows:Firstly, according to the requirement of research in this thesis, EOG collection experiments are reasonably designed, and the existing EOG databases are updated and improved for subsequent work.Secondly, in order to recognize and classify the EOG signals generated by typical eye movements (up-saccade, down-saccade, left-saccade, right-saccade and blink), the common spatial pattern (CSP) algorithm is employed to extract features of EOG waveforms. Combing with support vector machine (SVM) classifier, the EOG features corresponding to different eye movement are identified and classified, and high recognition rate of eye movement has obtained.Thirdly, the sliding window based cumulative recursive algorithm is applied to EOG endpoint detection and extraction of effective EOG aiming at the realization of EOG-HCI online system. The experiment of continuous EOG reveals that this algorithm can realize quick detection of effective EOG segments, greatly improving the accuracy in EOG endpoint detection.Finally, on the basis of EOG signal processing and pattern recognition algorithm, the online experiment platform of EOG-HCI is built and tested. The simulation experiments results indicate that this system can well identify different eye movement patterns to realize the online control of external devices (mobile phone).