為了進行自動化地從視訊資料中萃取出移動物體的軌跡與路徑線，本研究的視訊資料是利用固定方位與視角的攝影機，以一至兩台正在移動的機車或汽車為目標，由高處拍攝地面取得。利用以下3步驟對視訊資料進行軌跡與路徑線的萃取：1. 興趣視訊段落萃取：利用背景影像將減法快速偵測出視訊資料中含有移動物體的興趣視訊段落，並且將其萃取出來，以利後續的軌跡萃取與路徑線萃取等步驟；2. 軌跡萃取：利用主成分分析（Principal Component Analysis，PCA）與區域增長法為主的步驟，從各個興趣視訊段落中萃取出移動物體的軌跡；3. 路徑線萃取：萃取出軌跡的中心線作為路徑線，並針對不同的路徑線型態對路徑線進行匹配，得到物體的移動路徑線與方向。測試影片包含了單一移動物體、平行軌跡雙物體、X型軌跡雙物體、Y型軌跡雙物體等不同型態的興趣視訊段落。三段影片的測試結果顯示，不論測試影片中的各興趣視訊段落的型態為上述何者，本研究都能夠從測試影片中將其萃取出來，並且從興趣視訊段落中萃取出移動物體軌跡，且利用軌跡更進一步萃取出物體移動的路徑線與移動方向。 In order to extract the track and route of moving object from the video token by a static camera (which means a camera with fixed position and view angel), there are three steps in our algorithm: (1) Interest Clip Extraction: an interested clip means the clip containing moving object .The main purpose of this step is using background subtraction method to detect moving object and extracting interest clips from video data rapidly. (2) Track Extraction: this step employs principal component analysis (PCA) on each interested clips to produce principal component (PC) images, and then uses the segmentation technique to extract track regions from each PC images.(3) Route Extraction: this step extracts the central route from each interested clips, and applies different route matching algorithm on different type of route. In the end, we extract the routes with the direction of each moving objects. The test data includes number of moving objects in each interested clip that contains: (1) An interested clip that has only single moving object. (2) An interested clip that has two moving objects move with parallel track. (3) An interested clip that has two moving objects move with X-shaped track. (4) An interested clip that has two moving objects move with Y-shaped track. The experimental result indicates that our algorithm is able to reach the goal of extracting the track and route of moving object from each set of video data.