Kingshuk Chakravarty, Brojeshwar Bhowmick, Aniruddha Sinha, Hrishikesh Kumar, and Abhijit Das
Gait, Balance, Locomotion, Telehealth, Kinect, Curvature Analysis
This paper presents a Kinect based system with a novel approach to find human gait parameters. This system has huge impact in rehabilitation where patient's gait pattern analysis is monitored after stroke. Our method is the first of its own kind where gait as well as postural stability can be monitored using a single system, and can be used at homes by patients. This system can act as an affordable alternative to the costly GAITRite which is a standard gait monitoring device used in clinical practices. We use temporal skeleton information obtained from Kinect to evaluate all parameters. Segmenting the skeleton data in time is one of the key steps in our method. We employ robust curvature detection algorithm based on eigenvector decomposition, which works significantly better than standard second derivative based approach. The similar technology is also employed to understand the characteristics of patient's postural control in Single Limb Standing (SLS) exercise. Experimental results demonstrate that our method is comparable as well as compatible with the corresponding parameters of GAITRite and can also be used to calculate SLS duration reliably.