想要在動態影像中追蹤多目標移動物至今仍是一個重要且尚待解決的問題。在此篇論文中,我們針對背景會移動的影像序列,提出了一個有系統的方法來解決追蹤多目標移動物會碰到的問題。我們所提出的多目標移動物追蹤方法,主要由以下四個部份所構成︰多目標移動物偵測、物件關係之建立、背景影像的重建,以及各物件陰影的去除。為了有效的追蹤出多目標物件,我們利用影像畫面與初始背景影像的資訊及使用等高集合法(Level Set Method)偵測出多目標移動物,再利用編號演算法(Connected Component Analysis)和物件關係表來追蹤並記錄不同物件及其關係以達到追蹤的目的;為了解決背景移動的問題,我們利用區塊匹配(Block Matching)方式計算出整張影像的動態向量,並利用此資訊來補償背景影像,藉此解決背景變動時動態視訊物件追蹤所會碰到的問題,最後利用物件本身的動量(Moment)資訊配合物件的輪廓資訊來去除各物件的陰影。實驗結果證實我們所提出方法的優越性和穩定性。最後並討論此方法的限制性以及未來的相關應用。
This paper presents a novel approach to track multiple moving objects using the Level-set method. First, in order to overcome the problem of camera motion, a block matching method is used to estimate the information of camera motion for background compensation. Then, the difference image is obtained by subtracting two adjacent frames. With the difference image, different moving objects can be detected using the Level-set method. Furthermore, a Gaussian object model is used to remove object shadows. Finally, through connected component analysis, an object-relation table is created to track all the detected objects efficiently. Experimental results show that the proposed method is much robust and powerful than other traditional methods.