Individual Identification Using Lower Body Skeleton Joint Data from Kinect V2 Sensors

For having quite complex parameters for each human, gait in each individual becomes different. This characteristic that can distinguish the gait of each individual is used in the field of biometrics to identify the gait of the individual being monitored. This research aims to analyze a person's gait using the Kinect V2. The implementation of gait analysis was monitored by placing the Kinect V2 2.5 m next to the path of a person walking on a treadmill. We recorded subject walking using the Kinect V2 and the Kinect Software Development Kit (SDK) and produced skeleton tracking data. The data were then processed, including feature extraction using first-order statistics and data classification using K-Nearest Neighbors with the N-fold Cross Validation feature. The data classified in this study produced an accuracy of 60% -94% using the lower part joint of the human body. The results of this study indicate that the use of the lower body limb joint is sufficient to be used as a gait biometric with a simple method. The use of more advanced techniques will certainly guarantee higher accuracy.