由於近年來監控系統的硬體裝置與影像處理技術蓬勃發展,使得即時影像運算的時間大幅降低,吸引了各方的研究學者進行相關的研究與應用。監控系統如何快速地透過影像視覺判定並偵測出在工作空間中的人物,一直以來都是重要的研究課題。本研究使用背景相減法作為判定工作空間人物的法則,搭配運用簡單且運算快速的系統架構來進行系統的開發。系統架構包含三個部份,分別為影像擷取、影像前端處理與環境和人物判定。影像擷取部分利用網路攝影機webcam與接收工作空間的視訊影像;影像前端處理則使用C++並搭配OpenCV提供的函式庫來建構影像處理QT平台,將網路攝影機取得的影像資訊進行影像處理,利用背景相減法運算找出工作空間中人物影像與背景環境影像的區域,接著進行平滑濾波與影像二值化將移動的人物區域分割出來;在環境和人物判定部分,分析切割出來的移動人物的範圍並計算其像素點的變化量,判定是否為移動的人物。最後由系統呈現影像運算判定的結果,若有人物在工作空間中移動則立刻進行標示並顯示於螢幕上方,以達到即時監控系統的成效。
In the image processing research topics, the technology of how to instantly detect human body within a moving region has always been a hot research topic. In recent years, due to the rapid growth of the development in computer hardware, the detecting of the existence of human body via image processing systems has led to a significant decrease of time usage. In this research, QT platform combining with OpenCV library is applied to implement the detection of human body within an indoor moving region. The system contains three components, image capture, image pre-process, and environment determination. First of all, video image is captured via a webcam. Secondly, image pre-process is used to obtain moving pixel area by Background Subtraction in order to conduct smoothing filtering and image binarization. Finally, moving objects are determined by calculating the variations within an image moving region. The results whether moving objects in an indoor environment are detected will show on the left top corner of the screen.