Gait Based Personal Identification System Using Rotation Sensor

Abstract

Every individual has a unique walking pattern which may be used as a signature for identification. This paper describes the construction of a feasible human identification device using human gaits (Walking pattern). We have built a very low cost wearable suit mounted with eight rotation sensors, controller, software and power unit for measuring eight major joints of human body which are involved in locomotion. Different person’s walking patterns can be captured with this suit which has been named IGOD (Intelligent Gait Oscillation Detector). IGOD meets an excellent standard of accuracy for capturing movements of major joint’s oscillation during locomotion. A gait classifier has been designed using ANN (Artificial Neural Network) and LDA (Linear Discriminant Analysis) combined with a bottom up binary tree approach. The designed system is free from environment and projection uncertainties, which normally plague any vision based systems. The device has been tested with 30 subjects (Persons). We have achieved a high recognition accuracy of up to 100 % over this limited sample of 30 persons.