Other Parts Discussed in Thread: IWR6843, IWR1443
Hello everyone,
as part of my master thesis I am using the IWR6843 board to detect hand gestures.
I've planned using the IWR1443 to detect these and just use and modify the offered examples (swipe and twirl), but as it is not available with the old chip anymore, I had to switch.
So then I've used the "Multiple Gesture and Motion Detection" demo to detect the swipe gestures.
I've then tried to modify the script to detect twirl gestures as well.
I did so by following the description in the comments of the MSS main script.
By creating the crosscorrelation factor, caching the results for a couple of frames (about 10) and then deciding if it is a clockwise or anticlockwise direction.
This works more or less robust and should be fine for my use case so far. Unfortunately I do not have the time to add a neural network or so.
But regarding my written elaboration I would be very happy if you would be so kind to tell me some more facts about the contexts in signal processing.
Why does one calculate the crosscorrelation factor. What does it tell me? Why does one use these two features (weighted doppler and weighted azimuth)?
I do know the basics of signal processing, for example what a correlation factor tells me (value beeing between -1 and 1; to get information about the statistical relationship), but I don't have the details to write a chapter about it in my thesis explaining the reasons.
Thank you so much for your time and help in advance!
Best wishes