Badly need help!! KINECT

I need to know something on using kinect:

Can you please suggest and explain any algorithm that i might need here:
I will need to program these, I need an overview and a little specifics on:

Hand gesture detection using kinect(all letters of the alphabet + space)

Another algorithm for:

(after detecting all the letters) the computer shall speak the spelled word

Thanks!

For Kinect, get the SDK from microsoft or use the OpenKinect library. You can get sound, a colour image frame and a depth image frame (there are also some other features such as skeletonisation of the detected object).

For hand shape recognition, once you have obtained the segmented hand image (perhaps using a skin-colour seeded region-growing method), perhaps use one of the multitude of shape detection algorithms, e.g. learnt Fourier Descriptors (FD), Inner Distance (IDSC), multiscale convexity/concavity representation (MCC), Triangle-area representation (TAR), etc. There are many better methods for gesture recognition, this is just a very simple approach.

There are plenty of text-to-speech packages available; just choose one.

dougy83 said:

For Kinect, get the SDK from microsoft or use the OpenKinect library. You can get sound, a colour image frame and a depth image frame (there are also some other features such as skeletonisation of the detected object).

For hand shape recognition, once you have obtained the segmented hand image (perhaps using a skin-colour seeded region-growing method), perhaps use one of the multitude of shape detection algorithms, e.g. learnt Fourier Descriptors (FD), Inner Distance (IDSC), multiscale convexity/concavity representation (MCC), Triangle-area representation (TAR), etc.

There are plenty of text-to-speech packages available; just choose one.

Click to expand...

Thanks for those!

But I am a beginner when it comes to kinect. And my professor said something on using the Hidden markov method for the hand gesture.
What do you think? Which of those algorithms is better to use?

And can you please tell the step by step process on that specific algorithm? I urgently need this please. Thank you very much!

Interestingly my daughter attended a University lecture from one of the main the designers of the Kinect - as part of her course, which I found a little odd as she's a Chemist?.

Apparently it was very interesting though, and he was a very clever guy.

Can anyone show me the flow chart or how this project will work? I need to know what exactly the steps that will be performed.

I will appreciate the help. I just need the main points.

There's almost a step by step for HMM gesture recognition here https://www.researchgate.net/post/How_to_train_parameters_for_HMM_Gesture_Recognition10

It would be good if we could have a tutorial on here about hacking the kinect for begineers lol sounds intresting

this is for the methodology part of a research paper..

I don't know how this will be done though.

Can anyone help me with this?