A Kinect-Based Virtual Rehabilitation System through Gesture Recognition

Stroke is one of the most common causes of death and long-term disability worldwide. People with motor disabilities caused by stroke experience limitations in performing daily tasks independently, and traditional rehabilitation programs have tediousness and accessibility problems for the patients. It is essential to develop novel and convenient rehabilitation systems to help people with motor impairment overcome these deficiencies. In this paper, we present a Microsoft Kinect sensor (Kinect) based virtual rehabilitation system, which allows a human operator to transmit his gesture information to the virtual environment by performing specified actions designed for our verification tasks. We verify the effectiveness and applicability of the proposed system in two user studies. In the first study, we carried out a left hand and right hand dexterity comparison experiment, in which the characteristic data of normal individuals have been obtained as assessment criteria during their accomplishing designed tasks such as upper limb raising and elbow bending. In the second study, we presented a six-level recovery assessment criterion through a body parts motion monitoring based recovery assessment method. Results suggest that it is effective for our proposed virtual rehabilitation system to assist patients in conducting home-based rehabilitation without physician's supervision.