if i have two adcs(analog to digital converters) sampling a signal. Is it possible to synchronize the two adcs with deep learning? as far as i know, with super-resolution on image processing it is the same process. If you have multiple images and fuse them together, theoretically its the same as finding out the different sampling points of the signals(registration) and after that the reconstruction with interpolation. Therefore I ask myself if such a solution is known in signal processing to use a neuronal network to synchronize two signals. The signals I use are from accelerometers. I know, that they dont sample the same signal if i have acceleration in different axis. But if the vibration is on a rotating machine, there is a correlation between the signals on the different axes. So signals on the axes can be considered as different sampling points on a signal. if the question is unclear, pls give me hints and i will improve it.
Yes, the waveforms are 90° out of phase in theory, and yes, the speed can change over time. What I want, is to find the shift between the signals and then line them up. If the signals are aliased because of undersampling, the "line up" together is difficult. You can't just use a correlation because aliased signals are different although they sample the same signal (only out of phase). An easy line up is only possible, if they are not aliased. Not aliased signals would not improve the resolving power, because they don't contain information about higher resolution in the form of alias-effects. In super-resolution, you use aliased signals and try to find the offset between the signals(the shift/phaseshift). If you find them, you can interpolate the sample points and you can double or triple the resolution like using multiple adcs with synchronized sampling points.