Two intrinsic calibration questions

Home Forums AR Sandbox Forum Two intrinsic calibration questions

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
  • #103053

    Hi all-

    I’m attempting to do the intrinsic calibration of the Kinect cameras, and I’m running into two problems:

    1. The depth camera image is terrible. The edges of the checkerboard boxes are barely discernible, much less so than in the calibration video. The only obvious difference between the example semi-transparent grid and mine is that I used blue painter’s tape to make the squares instead of white. But depending on where I position the grid and camera, I can sometimes get a few of the edges to be decent which makes me think that color is not an important variable.

    2. If I ignore problem 1 and go ahead and fit the grids to the images, when I press the 2 key I always get “No valid depth plane to lock to.” According to the video it doesn’t seem as though there are any steps that I skipped, but I still can’t get past this step.

    Thanks very much.

    Westhill High School
    Syracuse, NY

    Oliver Kreylos

    I recommend skipping intrinsic calibration and per-pixel depth calibration on the first installation. The factory-supplied calibration downloaded in step 6 of the installation instructions is usually fine. If the alignment is more than a few millimeters off after projector calibration, I recommend doing the per-pixel calibration first. It’s pretty simple and the biggest cause of errors.

    But to answer your question: before aligning a grid select “Average Frames” from RawKinectViewer’s main menu and wait until the depth image no longer changes. Then line up the grids in the depth and color images and select a tie point. To go to the next tie point, un-check “Average Frames” again, move the camera, and repeat.

Viewing 2 posts - 1 through 2 (of 2 total)
  • You must be logged in to reply to this topic.

Comments are closed.