本研究提出利用圖框差異改善物件擷取正確性之方法。先根據錄影序列的内容使用疊代法建立背景模式。接著利用背景相減找出目前背景模式的相異區域作為粗略的 移動物件。再由圖框差異補償細微的局部資訊。應用邊緣跟隨法可以將擷取物件的不連續區域重新連結，最後使用形態學操作將中空區域填滿以得到完整的物件。從實驗結果可以看到本文擷取的移動物件具有相當高的正確率。

This work uses frame difference to improve the accuracy of extracted objects. automatic extraction of moving objects in video surveillance. Pixels at all positions use such an iterative technique to establish the background. The original image is subtracted by this background to obtain a difference image. The noises resulted from the rapidly move can be removed by the differential frame. The modified edge-following scheme is employed to connect the discontinuous contours of objects resulted from low contrast, and empty areas of the closed contours are filled by the morphological operation to obtain the complete objects. As compared to the watershed method, the proposed method have the best performance of moving object extraction in in-door cluttered background video sequences. Therefore, the proposed method can be widely and effectively employed in various video applications.