Video stabilization (VS) is becoming a video enhancement technique that aims to remove unsteadiness and unexpected vibrations in video sequence and makes it to be stabilized and smoothed in order to acquire more pleasant and better output one. This di...
Video stabilization (VS) is becoming a video enhancement technique that aims to remove unsteadiness and unexpected vibrations in video sequence and makes it to be stabilized and smoothed in order to acquire more pleasant and better output one. This dissertation presents a digital video stabilization process using Kanade-Lucas-Tomasi algorithm for tracking good features between two continuous frames. Basing on these feature information, motion parameters are estimated by using RANSAC method. Continuously, the estimated motion parameters will be sent to the motion smoothing stage, where the unwanted motion (jitter) is eliminated. A moving adaptive alpha-trimmed filter window is adopted to obtain the smoothen motion. Finally, the image warping stage warps the current frame according to the smoothen parameters and generates the stabilized video sequence.