I am trying to classify signals of healthy and faulty states in a mechanical system. What would be the best way to transform the signal to images except spectrogram and scalogram. I am using Matlab as programming language.
Why would you want to convert signals into images. Just use the 1D signals as the input to a 1D convolutional neural network. The labels can be one-hot encoded for classification purposes.
Ill agree with Anoop. Spend you time improving your label space. For one dimensional signals I've had great success using LSTM neural networks. You can see some progress I've made here www.github.com/leeleavitt/peakdeepdetect