none
get Depth pixel Coordinate from Color pixel coordinate? RRS feed

  • Question

  • Is there a method that can achieve the inverse of GetColorPixelCoordinatesFromDepthPixel ? like "GetDepthPixelCoordinatesFromColorPixel "

    How could I get the corresponding coordinates into the depth map for a point once I know the coordinate of the same physical point into the color image?

    I am segmenting the color image and now i need the depth information of the features that i segment out of the color image.

    Any suggestions are appreciated.  

    Friday, August 5, 2011 5:04 PM

Answers

  • Unfortunately, Kinect SDK Beta does not provide an inverse of GetColorPixelCoordinatesFromDepthPixel. We've seen requests for this in the past, and are prioritizing this among other features, so hopefully we can help you out better in the future.

    The best I can suggest is for you for now is to have something in your application where you would take a snapshot in both color and depth images and then do the depth-to-color mapping for snapshot and try to 'calibrate' an inverse mapping function for your application based on what you see.

    Good luck and sorry about limited support.
    Eddy


    I'm here to help
    Saturday, August 6, 2011 2:32 AM

All replies

  • Hi I was wondering about the same thing... I have tried almost everything. This is what I did- I cropped the RGB image (basically ROI )to match the Depth image, I resized the RGB image and the Depth Image to 640X480. I tested it out on a WPF by the mousedown position event whcih I used to draw out red dots. It looked okay, but when fed it back to the Skeletontodepth almost every value was very off. My goal is to use the RGB data to track, then somehow corrolate the coordinates from the RGB to 3D space. Another thing that I will try is to do some camera calibration with kinect (Opencv) then "reverse engineer" the bitmapsource data -somehow. How my "suggestions" helped....
    Friday, August 5, 2011 6:12 PM
  • Unfortunately, Kinect SDK Beta does not provide an inverse of GetColorPixelCoordinatesFromDepthPixel. We've seen requests for this in the past, and are prioritizing this among other features, so hopefully we can help you out better in the future.

    The best I can suggest is for you for now is to have something in your application where you would take a snapshot in both color and depth images and then do the depth-to-color mapping for snapshot and try to 'calibrate' an inverse mapping function for your application based on what you see.

    Good luck and sorry about limited support.
    Eddy


    I'm here to help
    Saturday, August 6, 2011 2:32 AM