Merge two kinect output images into one - the output resolution RRS feed

  • General discussion

  • Let's assume we have two kinect sesors set up on the one plane, but spaced p meters from each other.

    I want to merge both frames from both sensors, assuming the output from the 2nd sensor simply overwrites pixels from the 1st sensor in the intersection field. 

    How can I determine the output resolution for the output merged frame in case I want to use skeleton frames or depth frames?

    Sunday, August 24, 2014 2:23 PM

All replies

  • This is not a simple scenario to solve for since each setup is going to be different. You will have to be precise with measurements and know the math to map the data between cameras. There is also image stitching techniques you may need to employ that would be beyond what you might find here. There are third party libraries that might help.

    This would greatly depend on how the cameras are oriented in relationship to each other and if there is any camera distortion that needs to be taken into account. The simplest way would be to setup both sensors an a straight line and manually manipulate them so there is a couple pixel overlap so the stitching is easier. Keep in mind that there is still camera distortion that may affect what you are doing.

    Carmine Sirignano - MSFT

    Monday, August 25, 2014 7:54 PM
  • Hi, thanks for your reply. I'm assuming the simplest scenario here, where both sensors are set on a straight line, so that there is no deviation angles and differences on y and z axes. Any ideas?

    Sunday, August 31, 2014 10:23 AM
  • Anyone has any idea?
    Monday, September 8, 2014 6:33 PM