Detecting Key Inter-Joint Distances and Anthropometry Effects for Static Gestures Development using Optical Motion Tracking

Abstract

In the effort to design three-dimensional gestural interfaces for surface ship platforms, this study was conducted to evaluate the critical anthropometric factors that influence developing mutually exclusive static posture recognition algorithms. Using the Microsoft Kinect, 400 inter-joint distances determining critical key joint factors were collected to identify 15 static postures. By using a discriminant analysis, 9 out of 400 possible key inter-joint distances were identified for distinguishing the 15 postures. Comparing the discriminant function model predictions to the actual observed values for the 15 postures resulted in an overall prediction accuracy of 97 percent.