桌面顯示系統之手指偵測與手勢辨識

由於多指觸控科技的發展以及人性化操作方式的訴求，使得很多成熟產品的出現，其中互動桌面系統已經受到相當的注目，許多研究機構更致力於大型互動顯示系統的開發。基於此需求，利用紅外線攝影機及紅外線投燈器，搭配影像處理技術來實現手指偵測與追蹤的系統實為一個可行的解決方案。為了避免因為大量影像分析所造成的系統延遲，本研究中開發了一個運用圖形處理器做加速運算的高效能多重觸控手指偵測系統。同時，於影像處理的過程中，擷取的影像易受到打光不均勻的影響，而呈現亮暗不均勻的現象，因此對於影響系統穩定性的參數需要做調整，我們實作了一個基於亮度的適應性演算法，來達到效能上的期許。在手勢辨識方面，我們提出了一個軟體架構，能夠支援所有符合輸入格式的硬體裝置，並輕易地使用內建既有的手勢辨識演算法，來達成應用程式的快速開發，同時具備手勢的擴充性與跨平台及不同語言開發的能力。另一部分，我們嘗試讓使用者自行定義手勢，讓系統能夠學習並加以辨識，以期可以提升操作的便利性。

關鍵字

手指偵測；多重觸控；互動桌面；圖形處理器；手勢辨識

並列摘要

Due to the development of multi-touch technology and the demand for user-friendly interactive operation, there have been a number of successful products. Among them, interactive tabletop system is a remarkable one, and it is still in progress. Based on computer vision approach, it is capable of expanding to large scale with low cost. However, the computational efficiency would be a concern as the image processing loading increases. In order to overcome this issue, we develop a GPU-accelerated multi-touch detection system. In the meantime, we figure out the input images would have uneven brightness due to lighting. So we implement an adaptive threshold algorithm so that the system performance would be more preferable. As for gesture recognition, we propose a software architecture which can support a variety of multi-touch hardware, provide gesture processing, be cross platform compatible, and allow applications to be developed in the desired programming language. Besides, a prototype is implemented for the purpose of recognize user-defined gestures.

並列關鍵字

Finger Detection ； Multi-Touch ； Interactive Tabletop System ； Graphics Processing Unit ； Gesture Recognition

參考文獻

[2] Han, J.Y., “Low-Cost Multi-touch Sensing Through Frustrated Total Internal Reflection,” Proc. UIST '05, ACM Press, pp. 115-118, 2005.

[3] “Touchlib – A Multi-Touch Development Kit.” Available at:

[10] “Microsoft DirectShow.” Available at:

[14] N. Funk. A study of the Kalman ﬁlter applied to visual tracking. Technical report, University of Alberta, 2003.

[15] Intel. “Open Source Computer Vision Library (OpenCV).” Available at:

國際替代計量

桌面顯示系統之手指偵測與手勢辨識

全文下載

主題瀏覽