| Summary: | 碩士 === 國立臺灣科技大學 === 電子工程系 === 102 === Nowadays, the applications of motion-sensing technology have been growing up quickly. With motion-sensing technology, human can do more convenience and interesting applications without traditional controller. In developing motion-sensing applications, a multifarious design for every kinds of action control. Besides, it is difficult to design an overall action recognition method due to the different skeleton structure between each human. In this thesis, an action recognition system is provided by using information of human skeletons obtained from RGB-D camera. The region space is classified and collected feature sequence for fourteen actions via the coordinate transformation. Finally, the support vector machine is used to classify the feature sequence. The experimental result shows that the proposed system can recognize fourteen actions correctly.
|
|---|
