#1
  1. No Profile Picture
    Registered User
    Devshed Newbie (0 - 499 posts)

    Join Date
    Feb 2016
    Posts
    4
    Rep Power
    0

    Video stabilization using convex optimization


    Hello,

    I have posted this elsewhere but have never obtained any replies.

    I have been intruiged by the video stabilization done by YouTube and wanted to implement it on matlab. I tracked down two papers related to it, this being them,

    1.) Error 404 (Not Found)!!1

    2.) 302 Moved Temporarily (L1 L2 optimization for video stabilization)

    I've made some progress, but sadly all I can get is a skewed video. In working on this problem I encountered a lot of new things which I'm also trying to learn, but as I have no background in them, I'm facing a lot of difficulty.

    Is someone here well versed with such topics (Convex Optimization, CVX modeling in matlab, Homographies etc), if so, I would love if you could help me in solving this problem.

    I'm unable to get a proper understanding of the paper, I feel (?) that they may have omitted some details, but I'm not sure.

    Thanks.
  2. #2
  3. Contributing User
    Devshed God 1st Plane (5500 - 5999 posts)

    Join Date
    Aug 2011
    Posts
    5,792
    Rep Power
    508
    what is video stabilization? Is it moving frames a little bit to best match the previous frame?
    [code]Code tags[/code] are essential for python code and Makefiles!
  4. #3
  5. No Profile Picture
    Registered User
    Devshed Newbie (0 - 499 posts)

    Join Date
    Feb 2016
    Posts
    4
    Rep Power
    0
    "what is video stabilization? Is it moving frames a little bit to best match the previous frame?"

    Yes in a way, although there are a lot of algorithms out there they all usually have these three steps in common:

    1.) Motion Estimation: Estimating motion of the scene. Involves tracking of features of a landmark/background/reference (depending on approach and algorithm) present through the frames.

    2.) Camera Optimum Path: Called by various other names, the tracked features give polar transforms (homography or affine) which denote the image warp transform between image pairs. An optimum warp is calculated (this step varies from algorithm to algorithm)

    3.) Motion Correction: The warp/crop window transform / <suitable quantity as per your algorithm> is applied to fix the image. Sometimes the transform is simply translational, so as you said its like the camera accidentally shook left and up, hence the current frame would be warped (translated) to right-down by the needed measure. (Of course more complex movements like rotation would require affine transforms)
  6. #4
  7. No Profile Picture
    Registered User
    Devshed Newbie (0 - 499 posts)

    Join Date
    Feb 2016
    Posts
    4
    Rep Power
    0
    Key word here is and the most basic form of video stabilization entails Registration.

IMN logo majestic logo threadwatch logo seochat tools logo