Personal Identification by Pedestrians Behavior

Abstract

The recent progress of motion sensor system enables to the personal identification from the human behavior observed from the sensor. Kinect is a motion sensing input device developed by Microsoft for Xbox 360 and Xbox One. The personal identification using the Microsoft Kinect sensor, shortly Kinect, is presented in this study. The use of the Kinect estimates the pedestrian's body size and walk behavior. The human body sizes such as height, width and so on and the walking behavior such as joint angle, stride length and so on are taken as the explanatory variables. The models which identifies pedestrians from the explanatory variables are defined by the traditional neural network (NN) and Support Vector Machine (SVM). In the numerical experiments, the pedestrian's body sizes and walking behavior pictures are taken from fifteen examinees. The pedestrian's walking direction is specified as 0°, 90°, 180° and 225° and then, the accuracy was compared. The results show that the identification accuracy is the best in case of 180°-walking direction and that the accuracy of the support vector machine is better than that of the neural network.