Investigating webcam data capture

daccordeon / gravexplain

Continuous gravitational wave searches in a table-top experiment

BSD 3-Clause "New" or "Revised" License

1 stars 0 forks source link

Investigating webcam data capture #5

Closed daccordeon closed 4 years ago

daccordeon commented 4 years ago

take a test video from a webcam
load in data from webcam (direct or from video), write scripts to process to .csv or other format (e.g. can we just grab the green channel intensity and save frames of that)
make a time series out of a single pixel, then Fourier transform it (rings should be symmetric and all move at the input frequency, so taking a single point on a single ring should capture the motion)
estimate the noise level we can expect to see

daccordeon commented 4 years ago

I forgot that we're really just trying to track a point, and I know software for that! This is tracker, I'll see if it works, if it does and I don't have anything else I'll have a go at OpenCV stuff in python. https://physlets.org/tracker/

daccordeon commented 4 years ago

ran Tracker on prepare signal with period 10s captured at 30fps results were very clean, see tracker_signal.pdf tracker_signal.pdf

daccordeon commented 4 years ago

starting in on OpenCV, it's a great package! lots of fun, it directly provides bgr (like rgb) numpy arrays, so finding the green channel is a piece-of-cake (bottom-left):

opencvfun

daccordeon commented 4 years ago

https://pythonprogramming.net/loading-images-python-opencv-tutorial/

new goals:

try to track a very noisy video
see what frequencies are able to be tracked (expect 30fps to limit this) edit: grammar

daccordeon commented 4 years ago

noisy_tracker_signal.pdf

daccordeon commented 4 years ago

something to spend time on:

make a video of oscillating green rings, simulating the physics
add noise and track it

daccordeon commented 4 years ago

rings done

daccordeon commented 4 years ago

webcam works fine, despite the 480p grainy capture this should have been closed last week