Mirror Image

Mostly AR and Stuff

Marker vs markerless (bundle adjustment)

Here is a sample of image registration with fiduciary marker (actually the marker I used in my games) vs registration with bundle adjustment. Blue lines are points heights (relatively to marker plane) calculated using marker registration and triangulation. White lines are the same using bundle adjustment(modified). Points extracted with multiscale FAST and fitted with log-polar Fourier descriptors for correspondence (actually SURF descriptor produce the same correspondence).
marker vs markerless
As you can see markerless is in no way worse then markers, at least on this example ))).

23, July, 2009 - Posted by | Coding AR | , , , , , ,


  1. Looks like a very nice example!

    >Blue lines are … calculated using marker registration and triangulation

    Do you mean, that you in fact selfcalibrate cameras (views) with marker registration and triangulation and that it is based on apriori knowledges about marker ?
    I understand the difference between marker vs markless approaches, my question is only for clarification.
    Thank you.

    Comment by Petrov Alexander | 23, July, 2009

  2. About marker-based approach – I didn’t selfcalibrate camera. Camera parameter (focus distance) was measured separately. No noticeable radial or affine distortion in this camera. Marker was registered with least square shape constrain(could have been simple crossXcross products in fact, for single marker there is no big difference). Geometry of the marker was known a priory. After markers registered for both frames, points were triangulated using camera positions.
    For markerless approach only a priory knowledge was camera focus distance.

    Comment by mirror2image | 24, July, 2009

Sorry, the comment form is closed at this time.

%d bloggers like this: