Kinect – Facts:
There is no point in talking about the Kinect from a technical point of view and describing the multitude of sensors and imaging cameras it contains. So, first allow me to start by explaining some ambiguous facts:
- Not every person who used the open source libraries and was able to capture the Kinect’s depth, RGB and IR stream is a Kinect hacker. There is only one hacker of this device and credit should be given in the right place. [Competition Winner]
- There is no need to re-explain how the Kinect works, these folks explain it better [How it works]
- The Kinect’s full potential isn’t yet exploited; the open source libraries currently online don’t provide access to all of the Kinect’s capabilities.
- Microsoft is lame, and that is an undeniable fact.
Now, since the above has been explained, let’s get down to business.
Kinect – With Processing:
I have used a multitude of the Kinect’s open source drivers and language wrappers and the best wrapper with a lot of potential is the simple-openni for Processing.
Why? Because, processing is quite reputable, it has a small learning curve and it’s becoming quite reputable. Amazing stuff can be done with this language as it allows you to interface with a multitude of other digital devices as well.
You can download this wrapper from: Download the drivers come bundled as well; so install them and you’re good to go (it works all operating systems as well)!
Fingertip extraction with OpenCV’s blob detection v0.0.1
This small project was quite fun, simply because fingertip extraction hasn’t been done this way before. It is quite easier to do it in different language since the OpenCV library is quite extensive in C++ and not so many functions and algorithms are implemented for its Processing sister library. However it has enough to get you started.
You can download the openCV library for processing from here: Download
Here’s what I have so far:You forgot a username for the Vimeo shortcode
Below is a screenshot explaining how the detection works:
It uses both the simple-openni library for hand and motion detection as well as the OpenCV library for blob detection. It’s a combination of both libraries and detection mechanisms to have a stable hand ROI from which I can extract the blobs defining each fingertip. It’s not very stable yet and requires a good amount of work so I’ll be releasing the algorithm once I have something solidrobust going on.