MotionMan Code Request (from Video 7 of Programming Kinect for Windows v2 video series)

Question

Casey Meekhof: thank you for the great presentation from Video 7 of the Programming for Kinect video series
that was published back in July. I was wonder if there was anyway that you could post the code you wrote for the MotionMan program you presented in the video.  I am a fairly novice programmer and was attempting to replicate some of the noise/background
reduction and pixel filtering but am unable to do so with any sort of sophisticated coding (mine seems to eat up much more memory with rather poor results). 
I am hoping that this question reaches you as I noticed someone else ask a question in a similar manner
and you ended up responding through their thread.  
Thank you in advance.

Answer 1

Thanks for the reply Carmine. That is unfortunate that the code isn't available but perhaps you (or anyone else) can guide me in a right direction.  First, let me explain what I'm trying to do:
Essentially, I want to be able to detect motion from some initial frame capture for a body that is supposed to be stationary.  My current setup involves a simple button which takes in the first 10 depth frames and averages them together to create a
"mask" which will then be referenced during the motion tracking. Then, the display will show (over the color image of the mask) where motion has occurred by changing pixels to some color.
The program I have written is pretty novice and I don't believe it does buffering in the way that you describe (perhaps you can elaborate further on that for me as I'm unsure how to incorporate that). For cleaning up the image, I have simple "for"
loops that take out any 0 depth pixels and replaces them with the average of pixels around it. Additionally, I'm trying to remove any "jumpy" pixels (as described in the video) by making comparisons of that pixel location in frames before and after
it and replacing the nonsensical depth pixel with an average of the other two. 
For the comparison of the mask to the live feed, I'm simply creating 5 arrays (1 for each of 5 sequential frames), averaging them, and then comparing them to the mask.  I then move the data in each array to the next one down the line and bring in the
newest frame to the array set and repeat. 
These processes seem a bit rudimentary to me but I do not know a lot of advanced programming especially when it comes to the Kinect and C#.  Any advice you could give would be appreciated as my current methods are not producing accurate results.