Distinguishing tip-toe walking from normal walking using skeleton data gathered by 3D sensors

In this paper, we propose a markerless based approach to differentiate between two different human walking gaits, specifically, the normal walking versus tip toe walking. The markerless based approaches are suitable in many applications, such as screening of autistic children from normal ones based on their body movement patterns or detecting walking irregularities. We use a 3D Sensor to collect 3D skeletal data of subjects that represents their behavioral spatial data. Several features have been extracted and feature selection methods were used to determine the best subset of these features. The proposed classification approach has been implemented and tested on a group of 75 people reaching 86% CCR between tip toe walking and normal walking.