我們提出了一個新的行為分析的演算法，這個演算法是利用mean-shift顏色分群以及動量估計所結合而成的。Mean-shift是利用來在一段影片中，物件的追蹤以及擷取上。而動量估計可以幫助我們進行物件追蹤及物件的動作分析。當我們擷取出一整段影片的物件之後，會使用細化演算法去取得物件的骨架，並以此骨架為基準去進行物件的行為分析，透過動量估計演算法去針對前後兩張的物件進行分析之後，我們可以得到一個位於兩連續動作中間的動作之骨架模型。之後可以利用此新產生出之骨架模型進行新動作的再生。 We propose a novel motion analysis algorithm by using the mean-shift segmentation and motion estimation technique. Mean shift algorithm is frequently used to extract objects from video according to its efficiency and robustness of non-rigid object tracking. For diminishing the computational complexity in searching process, an efficient block matching algorithm: cross-diamond-hexagonal search algorithm was used. In the motion analysis procedure, the stick figure of object obtained by thinning process is treated as guidance to gather the statistics of motion information. The experimental results show that the proposed method provides precise description of the behavior of object in several video sequences.
