Thursday, March 21, 2013

This Is Interesting

I downloaded ImageJ and have been playing around with combining the 3D data into a composite stack. What's nice is that it will take several hundred individual scans and combine them by median or mean depth values. So of course I had to run the color-coded sediment and take a Kinect scan:

The process needs refining, but this might be exactly what I'm looking for...


  1. Very cool. Is the color the Kinect's RGB data? What is the vertical scale?

  2. That is the original Kinect RGB data mapped onto the depth. The vertical scale is in whatever units the Kinect uses to measure distance (need to calibrate for measuring millimeters).

  3. Hi Matt,

    How did you create the stacks? I cannot seem to get it to work with the ppm and pgm formats. I imported them as RAW. I can see the image fine when I just open it, but when I try to import the stack it looks funky.

  4. My process for looking at the data was to open a batch of images (drag/drop onto the application icon with the Mac) - ImageJ handles about 100 images. Then under one of the drop-down menus there's an option to "load images into stack." Under the "stack" options, I made a "Z-surface" based on the average intensity of the images in that stack.

    I used that averaged z-surface to render a 3D projection and then used one of the ppm RGB images as an overlay onto the surface. I should write up the workflow soon and post it.

    1. Matt, that worked fine. Still not sure why the "import" function didn't fare well, it must be parsing the ascii in an odd way.

      So if the sample rate is 30Hz and you have 100 images that is ~3 seconds of footage. I'm surprised that those anomalous dips on the right side of the image were not smoothed out. Are they real or just artifacts of the edge of the FOV?

  5. Jeff,
    Some of the errors are produced by the IR pattern reflecting off standing water in weird ways. I've expanded the description of my process in today's post