手勢在人類日常生活中扮演著一種非常重要的溝通管道,為了提升人類與電腦之間的互動,本研究提出一套「應用視訊之自動化手勢軌跡追蹤系統」。在本論文中,手勢軌跡主要是在手勢不變的前提下,以手指尖的位置變化路徑來定義。系統流程包括前置處理(Preprocessing)、手掌及手指分割(Segmentation of Palm & Fingers)、特徵擷取(Feature Extraction)、以及軌跡追蹤(Trajectory Tracking)等。系統核心技術牽涉以規則為基礎作為分割手掌及手指的方法、手勢特徵擷取、與手指尖軌跡追蹤的方法。經由研究結果顯示,本系統可以成功地分割人類手勢的手掌區域及手指區域,並且可在某特定手勢中追蹤單一手指或多個手指的軌跡,也可以在手勢變化後重新追蹤軌跡。總結而言,我們期望本研究所提出的系統可以使電腦更有效詮釋人類手勢運動所蘊含的意義,有助於後續運用視訊攝影機作為輸入設備之人機介面發展。此外,本系統也可能取代其他必須搭配特殊的運動感應器或標籤之運動擷取系統,相對提供合乎成本效益的解決方案。
Hand-pose is one of the most important communication tools in human’s daily life. To enhance the human-computer interaction, the objective of this research is to develop a system that automatically traces the hand-pose trajectory using video sequences. Here, the hand-pose trajectory is defined as the path of position changes of fingertips, with the assumption that the hand-pose remains invariant. Our system includes: preprocessing, segmentation of palm & fingers, feature extraction, and trajectory tracking. The core techniques involve a rule-based approach for the segmentation of palm and fingers, and methods to hand-pose feature extraction and fingertips’ trajectory tracking. The experimental results demonstrated that our system is able to successfully segment fingers from a human hand, to trace the trajectories of single or multiple fingertips given a specific hand-pose, and to re-trace the trajectories upon hand-pose changes. In conclusion, our system could ultimately be incorporated in a computer user-interface that uses a video camera as an input device, and interprets the hand-pose in motion. In addition, our system may offer a cost-effective solution to other motion capturing systems that often require special motion sensors or markers.