主動式多攝影機視訊監控系統之研究

Abstract

近年來安全監控的應用越來越廣泛，相關的研究也越來越多。在本計畫中， 將研究討論如何建立一主動式的多攝影機視訊監控系統。其中，多部攝影機可以 擴大監視的範圍，同時也可以提供更多的資訊以進行更有效且穩定地監控，而主 動式的攝影機，則可更清楚地監視目標的資訊，以達到更好的監控效果。在本計 畫中，我們希望能夠發展一套主動式多攝影機監控系統，這套系統將具有以下特 點: 1. 為一多攝影機之系統，以便提供多面像的監控視角。 2. 此系統將包括PTZ(Pan, Tilt, Zoom)攝影機，這些攝影機將可以透過某種控制機 制來進行左右轉動(Panning)、上下轉動(Tilting)以及前後縮放(Zooming)。 3. 此多攝影機監控系統中各個攝影機將透過某種協調機制來互相合作，以達到較 佳的監控效果。 這個系統不僅可以推算出被追蹤物體的三維軌跡，也可以估測這些物體的大略姿 態。基於這些對於被追蹤物體的了解，本系統將會控制協調各攝影機的拍攝角度及 鏡頭縮放，以達到較佳的視訊監控效果。

Recently, research works and applications about video surveillance are getting more and more popular. In this project, we aim to develop an intelligent video surveillance system equipped with multiple active cameras. The use of multiple cameras may greatly widen the scope of surveillance, while the use of active cameras can further track targets in a dynamic and flexible manner. In this project, we’ll develop a surveillance system that (1) contains multiple cameras to provide different monitored views; (2) contains active cameras to provide capabilities in panning, tilting, and zooming; and (3) contains a control mechanism that can coordinate cameras to provide dynamic and flexible video surveillance. This system may not only reconstruct the 3-dimensional trajectories of the monitored objects, but also roughly estimate the gesture of objects. Based on the understanding of the monitored objects, the system can automatically adjust the poses of cameras to achieve better performance in video surveillance.