Rozpoznávání a detekce pohybů pomocí senzoru Microsoft Kinect

Abstract

In this thesis is discussed current situation at field of person movement recognition using Microsof Kinect device and skeleton tracking. Situation is concretized on driver in car. Currently this area is not publicly known well. Most of applications for skeleton tracking using Kinect are optimized for standing person and environment where are not so many objects surrounding tracked person like steering wheel. This is the reason why current skeleton tracking applications fails. In this text correction of current methods is presented with help of static background removing and improved tracking algorithms for elbows and hands. As default library for skeleton tracking “Skeltrack” is used. OpenKinect drivers are used for communication with Kinect device.