Extraction of Stair Walking Parameters in Living Space by Using Kinect v2

Abstract

These days, many problems are associated with the increase of people need nursing care due to the aging society. To help the problems, expanding the healthy life expectancy is important. Thus, we have been suggesting the “Monitoring System” which can monitor the resident’s daily life. This system aims to monitor changes of resident’s behaviors as unusual signs to prevent related accidents or diseases. As a part of the system, we focused on walking. As it is reported that the ability of walking is related to many disabilities, analyzing the walking patterns is effective to know the aging level. Especially for stair walking, relatively higher ability is needed compared with level walking, that is, the decline of walking ability will appear in stair walking earlier. Therefore, we focused on the stair walking in this study. In the previous researches, sensors such as markers or electrodes are mainly used to obtain the walking information. As our system is intended to be used in living space, sensors attached to a subject’s body are unacceptable. Thus we used Kinect v2 to acquire the coordinate points of human joints without restraining subjects in this study. We use depth data considering the privacy protection. Several verification experiments to evaluate accuracy of our proposed system have been conducted.