在這個電腦網路教學的時代裡，技術與設備汰換速度相當的快。不單是遠距教學，連同視訊會議，亦高度依賴著電腦技術及其支援設備的發展。多媒體教材(Content)與演講者影像(People)一併傳送到接收者眼前已經變成了不可避免的趨勢。然而，能達成此目標的設備價位相當高，也因此阻礙了實際應用。倘若設備的價格能降低，同時又能維持同等的品質與功能，則毫無疑問地，這類型設備的應用將會被廣泛的快速推廣。 本論文在發展一套可自動合成演講者影像及教材的系統，以達到較有效率的教學目標。在本系統裡，我們將結合演講者的影像，並將之混合於演講內容最適合的位置，一併呈現給接收端，採用的都是以影像為基礎的技術。 實驗部份，透過許多影像合成的測試，證實出我們提議的系統之可行性及有效性。實驗結果顯示此影像及內容自動結合的系統確實可行。In this era of computer network teaching, the speed of technology and equipment replacement is very quick. Not only distant learning but also video conferencing rely heavily on the development of computer technologies and supporting equipments. It is an inevitable trend that theimage of speaker in conjunction with the speaking content can be delivered to the receiver simultaneously to increase the learning effect. However, the equipment in achieving the goal is quite high to hinder itspractical applications. If the cost of the equipment can be reduced while attaining its original functionality and quality, it will surely boost itsspreading. In this thesis, we aim at developing a system which canautomatically compose the images of speaker and speaking content to achieve the goal of effective teaching. In the proposed system, the image of speaker will be captured and pasted on the best-fit part of speaking content to form a composed image to be sent to the receiver. The techniques that we adopt in accomplishing the goal are all image-based technologies. In the experiments, various video clips were tested to verify the feasibility and validity of our proposed system. Experimental results reveal the soundness of the proposed system in automatic speaker andspeaking content composition.
