none
Image Segmentation RRS feed

  • Question

  • My purpose is segment the video stream.

    For example, I want to show the RGB image that their depth value are smaller than 1m.

    How do I get it?

    By the way, when I use the method that called GetColorPixelCoordinatesFromDepthPixel,

    should I have to stand in front of kinect then I get the pixel coordinates?

    If I don't do that, skeleton engine won't tracks me and I can't get  pixel coordinates?

    I'm confuse by this issue.

    Any suggestions are appreciated.

     
    Tuesday, February 28, 2012 3:28 PM

Answers

  • If you want to show RGB where Depth is < 1m:

    When you receive a depth frame, call GetColorPixelCoordinatesFromDepthPixel when for each pixel where depth was < 1m.

    The output of this function will be the X&Y in the color image. 

    You could then set the corresponding pixel in a mask. 

    Then draw the color masked off by the mask.

    You do not need to stand in front of the kinect or to have the skeletal engine turned on. 

    The depth and color cameras are offset from each other, angled differently (to help compensate for being offset), and have different frusta.

    GetColorPixelCoordinatesFromDepthPixel just maps an X,Y,depth in depth space to an X,Y in color space.

    • Marked as answer by Berry Huang Wednesday, February 29, 2012 1:49 PM
    Tuesday, February 28, 2012 4:37 PM

All replies

  • If you want to show RGB where Depth is < 1m:

    When you receive a depth frame, call GetColorPixelCoordinatesFromDepthPixel when for each pixel where depth was < 1m.

    The output of this function will be the X&Y in the color image. 

    You could then set the corresponding pixel in a mask. 

    Then draw the color masked off by the mask.

    You do not need to stand in front of the kinect or to have the skeletal engine turned on. 

    The depth and color cameras are offset from each other, angled differently (to help compensate for being offset), and have different frusta.

    GetColorPixelCoordinatesFromDepthPixel just maps an X,Y,depth in depth space to an X,Y in color space.

    • Marked as answer by Berry Huang Wednesday, February 29, 2012 1:49 PM
    Tuesday, February 28, 2012 4:37 PM
  • I see.

    Thanks for your advise, I know all of the story and I'll try to implement it.

    Wednesday, February 29, 2012 1:49 PM