New Methods for converting signal to image

Hi, Does anyone know any new methods for converting signal to image rather than STFT or Wavelet?
if there is any article or implementation of the code I will appreciate your help!

What kind of image are you thinking of?

I have LFP Data from Alzheimer’s mouses. I want to convert the signal to a time-frequency image for analysis using deep learning models like capsule networks. But I want to get some perspective about novel methods to convert the signal to the image to make a new method. Some ideas like for example making the full width at half maximum or wavelet frequency range, trainable in the network while converting signal to image using wavelet and convolution.

Do you mean using DL to run the time-frequency analysis? I don’t think I’d recommend that. Time-frequency analysis is a simple linear transformation, which is not something that DL is useful for. I think it would be better to use standard methods for the time-frequency decomposition, and then let a CNN learn the features of the time-frequency maps that best distinguish the Alzheimer’s from control data.

1 Like

Thanks Mike I solved the problem as it was for last month :wink: