電腦視覺在移動平台上的應用日趨發達,其中物件追蹤在手機等移動平台的應用中扮演重要角色。在移動平台上追蹤物件困難點在於,畫面中不僅有目標物件在移動,同時背景也不停變化,物件可能出現的範圍比靜止平台的物件追蹤要大上許多。當物件在前後畫面位置差距大時,對於物件移動的預測就顯得格外重要。因此,本論文提出適用在移動平台追蹤上的運動補償模型,透過粒子濾波器(PF),估測出補償相機移動量後之物件在二維影像中的運動量。粒子濾波器利用相機和物件的補償相機移動量後之物件在二維影像中的運動量,估測物件在二維影像上的運動量。相機的運動量,則使用SURF演算法計算。實驗結果顯示,我們提出的追蹤演算法應用在移動式平台上時,對於移動快速或不規則運動的物件皆能有良好的追蹤效果。Robustness of visual tracking on mobile devices plays a key role in the success of emerging applications of robotics and cell phones. The prediction accuracy of object motion makes a great impact on the tracking accuracy. Therefore, this paper proposes to combine motion compensation and motion model to achieve fast and accurate estimation of object motion in 2-D images. At each time instant, the particle filter estimates the 2D object motion after motion compensation in the image based on the information of camera motionCamera motion is extracted with the aid of Speed-Up Robust Feature (SURF) algorithms. With our tracking algorithm, accurate tracking on mobile platforms can be achieved with a small number of particles. The experimental results show that our proposed tracking algorithm on mobile platforms performs well even if the rapid and irregular object motion exists.