A new feature extraction technique for person identification

Abstract

This paper proposes a one of a kind model of human distinguishing proof by which we can supplant the persevering work of individuals if there should arise an occurrence of security. Our technique is for the most part in view of human structure identification and face detection with Microsoft Kinect V1.0. With a specific end goal to apply our approach, 6 points of 20 in body structure and 26 points out of 121 in face detection is taken as input. These points are X, Y, Z coordinates outputted by Kinect skeletal and facial following output by Kinect skeletal and facial tracking. 16 unique distances are then calculated by the Euclidean distance formula using the coordinates. These are the selective components extracted from each user and afterward put in a database. At long last, by coordinating calculation our framework distinguishes known or obscure users progressively. Final output is then given as a result with aptitude and noteworthy precision.