chiragsalian Goto Github PK

big_data_project

javascriptexercises

Basic javascript exercises for beginners

motion_stabilization_using_ransac

We present a technique of video stabilization to remove out external disturbances caused by shaky hands, turbulence etc. Our technique involves in used Matlab video processing toolkit to process the video. We can then break up the video into frames. By doing so we can move on to process the frames and extract key robust feature trajectory points for each frame. Once we have the key feature points for each frame, we take these feature points from two frames and calculate a high dimensional distance vectors for these points by applying a feature matching technique. Based on a certain threshold we choose the distance vectors with the lowest values. We proceed to apply RANSAC algorithm to find the most appropriate distance vectors without skewing our results by not including the outliers. By using this data we can apply the geometric transformations on our frames to obtain a more stabilized video. We present many examples demonstrating the effectiveness of our approach to stabilize the video as efficiently as possible.

opengl