A method of rehabilitation training for finger pair movement based on multi — Leap motions

This Finger Pair Movement (FPM) is the key indicator for Stroke recovery. This paper presents a multi-Leap Motion-based training method for Finger Pair Movement (FPM) rehabilitation. This method first uses 3 Leap Motions to obtain the hand movement information and uses the Iterative Closest Points Algorithm (ICP) to register finger movements precisely from different angles, which could avoid the traditional obscured problem from the single angle. Then use the Self-Organizing Maps algorithm (SOM) to map the required 9 Finger Pair Movements. ICP algorithm obtains 3D point cloud from multiLeap Motions and completes data registration, which outputs high integrity three-dimensional spatial information of the hand joint skeleton. Using the SOM algorithm can solve the problem of obtaining data containing too much noise. The proposed method has been employed in clinical trials. The experimental results show that the proposed method has high efficiency, high error tolerance, and good performance.