I dont project about process realtime data from cyber glove( Virtual Hand ) . So I need to write some application that get realtime data from glove and feed to some algorithm. I don't know how to deal with process realtime data; does anybody have some resources?
1 回答
I'm pretty sure the cyber glove you're using comes with an SDK as well as examples on how to get the data from the device.
From there, I'm afraid we can't tell you much. I see you tagged your question with "recognition" but what are you trying to recognize exactly?
Recognizing gestures would typically mean analyzing a trajectory in 3d space.
I've never worked with such a glove but I can imagine it streams a sequence of data the same way a Wacom tablet would stream a sequence of (x,y, pressure) and eventually proximity and pen tilt data.
So, you will need to extract pertinent features out of this raw data in order to form what's commonly called a "feature vector". For instance you could resample the data using an interpolation scheme to end up with n tuples, each tuple containing information such as:
- position
- orientation
- velocity
- acceleration
- curvature
- etc
You will have to experiment in order to decide which features are best for the problem you're trying to solve.
Once you are able to convert a raw 3d trajectory into a normalized feature vector you will need to make a decision about the method you want to use, for instance:
- example based analysis with a DTW (Dynamic Time Warping) approach
- neural network training
- support vector machines
- there are many!!!
Unfortunately, pattern recognition is a vast subject and I can't tell everything about it in such a short answer. It's now up to you to study the literature.
Good luck.