基於人體姿態估測之投影片演示控制系統設計

The Design of Slide Presentation Control System Based on Human Pose Estimation

葉明揚(Ming-Yang Ye)

指導教授 ： 林昇甫

國立交通大學/電機學院/電機與控制工程系所/碩士(2014年)

https://doi.org/10.6842/NCTU.2014.01047

摘要

近年來，隨著手勢辨識研究的演進，許多利用體感操控之應用也逐漸發展，使得人們只需站在攝影機前便可操控電腦。有鑑於在會議室中之投影片演示時，演示者無法適時地站在投影布幕前利用雙手指示重點，而只能侷限於在電腦前演示投影片。
本研究藉由微軟推出的Kinect，應用手勢辨識於控制PowerPoint演示。本論文透過擷取人體參考點，辨識雙手相對於人體之單一方向動作，希望達成包含能夠讓使用者順暢切換頁面、在一定程度上的展示圖片功能及簡單的畫面指示效果功能之投影片控制輔助系統。
本論文之貢獻有兩點：第一，本論文提出一手勢辨識系統，在不同體型及臂展之成年使用者間，皆能達到一定程度之辨識效果。第二，本論文之系統整合圖片展示功能及畫面重點指示於投影片系統中，讓使用者能夠容易將投影片中之內容演示。

關鍵字

KINECT ； 體感操控 ； 手勢辨識

並列摘要

Recently, as the evolution of the researches on hand gesture recognition, many somatosensory control applications are gradually developed, and so that people can stand in front of the camera to control the computer. In view of the slide presentation in the conference meeting room, the presenter is limited in front of the computer, and couldn’t use their hands to point out the emphasis on the projection screen in time.
This thesis describes a gesture recognition system which is used to control a PowerPoint presentation by the Microsoft Kinect. This system focuses on the hand gesture recognition by extracting the human body reference points and recognizing the gestures which the hand movements are one direction related to human body, and then make the users to navigate the slides as fluid as possible, to resize an image on a slide and to display a pointer on slides which they want to mark to the audience.
The following are the contributions in this thesis. First, we proposed a gesture recognition system, which can achieve a limited extent recognition effect on adults. Second, it integrates an image-showing function and a pointer function in the system, making presenter easily show the key point they want to display on a slide.

並列關鍵字

KINECT ； depth camera ； human pose estimation ； gesture recognition

參考文獻

[1] M. Camplani and L. Salgado, “Efficient Spatio-temporal Hole Filling Strategy for Kinect Depth Maps,” Three-Dimensional Image Processing (3DIP) and Applications II, vol. 8290, pp. 82900E, 2012.

[2] K. R. Vijayanagar, M. Loghman, and J. Kim, “Real-Time Refinement of Kinect Depth Maps using Multi-Resolution Anisotropic Diffusion,” Mobile Networks and Applications, vol. 19, no. 3, pp. 414-425, 2014.

[4] P. Henry, M. Krainin, E. Herbst, X. Ren and D. Fox, “RGB-D Mapping: Using Kinect-style Depth Cameras for Dense 3D Modeling of Indoor Environments,” International Journal of Robotics Research, vol. 31, no. 5, pp. 647-663, 2012.

[7] I. Oikonomidis, N. Kyriazis, and A. Argyros, “Efficient Model-based 3D Tracking of Hand Articulations Using Kinect,” in Proceeding of BMVC, pp. 101.1-101.11, 2011.

[19] K. R. Konda, A. Königs, H. Schulz, and D. Schulz, “Real Time Interaction with Mobile Robots Using Hand Gestures,” in Proceedings of International Conference on Human-Robot Interaction, Boston, Massachusetts, USA, 2012.