Head Gesture-based Human-intelligent Wheelchair Accessibility Interaction

Recently unobstacle information technology takes more and more attention. It is mainly research how to help elderly and disabled people free to participate in social activities, and human-computer interaction based on the head gesture is a hot research field. So the research of human-intelligent wheelchair interaction based on the head gesture is a very interesting topic.This thesis researched the method of human-intelligent wheelchair interaction based on the head gesture, proposed the head gesture recognition based on lips detection, and use the recognition results to control the intelligent wheelchair. The article introduced the Adaboost(Adaptive Boosting)algorithm target detection, which use the weighted voting to generate the lip classifier cascades to detect the lips. But the accuracy of using the Adaboost algorithm to detect lips and identify head gesture is not high, and the time is to long. For these two problems, we proposed the hybrid Kalman filter algorithm with the Adaboost algorithm to detect lips. By predicting the location of the upper left corner of the lip in the next frame, reduced the lips detection range, improved the accuracy of the lips detection and head gesture recognition, and reduced the time.Finally, the thesis designed a head gesture-based human-intelligent wheelchair interaction system, and did a large number of experiments on this platform. The experiment results show that the reliability of the recognition algorithm of head gesture based on lips detection proposed by this thesis is quite Hexcellent H, the movement accuracy of intelligent wheelchair is more than 90% and the robustness of the intelligent wheelchair is effectively improved.