Application Research On Vision-based Augmented Reality Technology In Hand Function Rehabilitation

Augmented Reality is a new technology developed on the basis of virtual reality.It overlays the computer-generated virtual information on real scenes where users are and realizes augment of the real world. Users could conduct interaction with real and virtual items in the environment in a more natural way. Application of augmented reality in the field of rehabilitation has been a hot research topic at home and abroad.It could not only help patients train in a way that is closer to life, but also help patients be independent of therapists and do rehabilitation training independently,and it is more authentic.With National Natural Science Foundation of China as the background, and in combination of markers and natural features, this paper designs and develops a vision-based augmented reality system for hand rehabilitation training. The main research contents in this paper are as below:This paper makes correlative comparison of current status of research on marker-based augmented reality,and analyzes the design principles and deficiencies of various markers. According to the needs of hand rehabilitation training, this paper improves color markers with Hamming code for hand rehabilitation training and verifies the robustness and practicability of colorful markers through the experiment.Then this paper makes related researches on natural feature based augmented reality technology,and puts forward a tracking registration method based on ORB and KLT and combined with homography matrix. The innovation of this method lies in improvements on the traditional KLT algorithm by adding the image pyramid to analyze the tracking performance under circumstance of motions with large feature points, and using the improved KLT algorithm in combination with homography matrix to track and register target features. The feasibility of the method are verified through the experiments.Finally, this paper designs and develops a vision-based augmented reality system for hand rehabilitation training. Combining marker-based and natural feature-based tracking registration method, this paper designs three rehabilitation projects, namely pillar training, peg inserting training and Whac-A-Mole. These three project are progressively more difficult, which mainly exercise the functions of fingers, wrist joints, elbow joints and shoulder joints and so on. Finally, it conducts evaluation and analysis on questionnaires through the experiment, and verifies the effectiveness and practicability of the system.