機械人服務於社會與人類一直是機械學者們共同努力的目標,因此各種機械領域的研究人員都積極的投入研究,本論文也為此目標的達成,以視覺領域來盡上一份心力。 當Xbox在2010年正式推出後,Microsoft讓機械視覺的研究推升到一種新的層次,之後甚至將Kinect for Windows SDK不藏私地公布於世,並讓有興趣的開發者能盡情的發揮。這樣新穎的技術工具取代一般CCD,讓研發人員有了最棒的得力助手。 本論文是以C#來撰寫,使用WPF的開發環境,搭配上Microsoft於Kinect for Windows SDK 1.8版的開發資料庫,和從1.5版後所提供的Developer Toolkit,並且運用其公開的開放來源來撰寫本研究。 程式結構以深度串流和骨架資訊為基礎,擷取出人臉辨識所需的資料串流,以及手指基礎動作辨識的判定;運用人體關節點座標系統,自行開發姿態判斷,設計特定姿態以達成目標訊息,得以完成機械人與人體互動的訴求。
The application of robot on the human being life is the most popular topic that mechanicians keep studying. Hundreds of researchers have participated in this unlimited technology from many kinds of field that related to mechanics. This study focused on the interaction between robot and human being by the facial recognition technology. For accomplishing the purpose that the human being interacts with robot, this study developed a posture judgment by using human joints as the coordinate system. Combine the needs of data streams capturing from the result of facial recognition, and the recognition of basic finger movements to achieve the whole discrimination. This thesis is written in the “C#” computer language with the WPF development environment and Microsoft Kinect for Windows SDK rev. 1.8 as the development database.