The Data Optimization Approach For Leap Motion And A Study Of Hand Rehabilitation Training Game

Leap Motion,as a new type of hand interaction input device,is popular among researchers due to its small size and ease of use,but the accuracy of the hand motion data obtained is not high due to its susceptibility to self-masking and unstable sampling frequency.Unlike the fields of human-computer interaction and gesture recognition,the accuracy of the hand data acquired in the field of rehabilitation medicine has high requirements,and the level of accuracy directly affects the assessment of hand function and the designation of subsequent treatment plans.Therefore,this paper proposes a CNN-based data optimization method to improve Leap Motion data accuracy.In addition,to address the problems of traditional hand rehabilitation training,such as boring training and subjective evaluation criteria,this article uses Leap Motion with data optimization methods and gamification elements to design a hand function rehabilitation training system,which ensures the ease of use and reasonableness of the training system while meeting the accuracy requirements of hand function rehabilitation training.The research for this paper is as follows.(1)In order to meet the data needs of hand motion optimization by applying deep learning methods,this paper proposes a synchronous acquisition scheme for Leap Motion and motion capture devices,and constructs a multimodal synchronous data set to compensate for the shortcomings of the current hand motion data set which is not suitable for Leap Motion optimization.(2)Combined with synchronous data sets,this paper proposes a data optimization method based on CNN networks,which can conveniently improve the data accuracy of Leap Motion;at the same time,a finger plane constraint is proposed to make the network output more consistent with the finger kinematic characteristics in the network training process,combined with hand structure features.(3)By establishing a synchronized data set of hand movements for rehabilitation evaluation criteria and introducing a CNN network-based data optimization method.A hand rehabilitation training system based on Leap Motion was designed and implemented based on the Unity3 D engine,and finally the ease of use and reasonableness of the rehabilitation training system was experimentally proven.