Rejestracja ruchu za pomocą urządzenia Microsoft Kinect

Artykuł dotyczy tematyki rejestracji ruchu postaci ludzkiej (ang. Motion Capture). Praca pokazuje jak do rejestracji ruchu można wykorzystać kontroler ruchu Microsoft Kinect. Praca zawiera wyniki badań weryfikujących jakość oprogramowania NITE służącego do wyznaczania rotacji członków ciała w stawach. Badania wykazały, że jakość rejestrowanego ruchu nie jest wysoka. Mimo tego uzyskane dane mogą być wykorzystane do analizy gestów lub rozpoznawania ruchu.

---

This paper refers to a subject of a human body motion capture. Common video based motion capture systems are very expensive. This paper shows that it is possible to use a cheaper solution for motion capture which is a Microsoft Kinect movement controller. There is presented the software for motion capture of a human skeleton. We may use Microsoft SDK programmer toolkit or freeware frameworks like CL NUI Platform, OpenKinect and OpenNI for this purpose. We used OpenNI and NITE libraries (diagram in Fig. 2) for our research. The results of measurements of human joint angles with use of Kinect and our OpenNI based application are given and discussed. These angles are used for recording the movement into motion capture BVH files. We used free application Brekel Kinect (Fig. 3) for this purpose. The recorded data can be used for a realistic animation development. In our case, the data was used as an input source for our own visualization application made in Java language. The quality of recorded motion is not very high due to the calculation delays and problems with appropriate joint localization by the NITE software. The Kinect controller is not suitable for very accurate motion capture, e.g. for the needs of biomedicine, but the captured data are of sufficient quality for research related to the analysis of gestures and remote control.