Video stabilization algorithm using a moving alpha-trimmed mean filter window

In this paper, a fast video stabilization approach is presented to remove the unwanted camera vibration in video sequence. The proposed method first tracks good features from video sequence by using Kanade-Lucas-Tomasi method, then estimates the affine transform parameters using RANSAC algorithm. Each consecutive frame´s parameters are summed up to form the accumulated trajectory and a moving alpha-trimmed mean filter is used to achieve the smoothed motion. Experimental results show the capability of the proposed algorithm.