Contact

Dr. Vijayan Asari, Director

Phone: 937-229-4504
More
Vision Lab

Video Stabilization

Scale Invariant Feature Transform Based Video Stabilization

Video stabilization is a video processing technique used to eliminate the shakiness in video. The video captured from a handheld device or an autonomous robot traversing a rough terrain is shaky because of undesired movement of the camera. It is very difficult to track any objects of interest or extract details from the scene under such circumstances. The objective of this research is to eliminate the effect of high-frequency motion of the camera which causes the shakiness in video. The video stabilization algorithm includes the following steps:  feature extraction, motion estimation and smoothing, and motion compensation. Feature points in each frame are tracked over successive frames to estimate the motion of camera. Scale Invariant Feature Transform (SIFT) is used for feature extraction. The camera path is smoothed using a filter, and the frames compensate for the motion and stabilize video as output. The amount of required smoothing is estimated based on the estimated camera motion.

Algorithm Flowchart Video Stabilization

Video Demonstration