Getting the Proper Feature Vectors from the Hilbert- Huang Transform?

I am currently in the process of coding up the Hilbert-Huang Transform in MATLAB for extracting features from infrasound signals. Fundamentally I believe I understand the whole process, but I may be missing something in the actual algorithm implementation. You take the Empirical Mode Decomposition and then apply the Hilbert Transform to the resulting IMFs. Then I think you take the imaginary values from the resulting complex numbers and those are your feature vectors, but that part I am less certain about, because I feel like there is a step missing after that. (Like taking the Power Spectral Density or something similar.)

My question is: At the end of the Hilbert-Huang Transform, what are the feature vectors? I thought they might have been the imaginary values of the complex numbers that result from the Hilbert Transform, but I am less certain of this now. I have also seen one or two people take only the first few IMFs from the signal as their feature vectors, which makes sense, but I am uncertain on how to determine this amount. (Unless this is decided by the programmer.)

I also apologize if this is more obvious than I am thinking. I am still learning about feature extraction and infrasound, so this is all still pretty new to me. All of the documents and sites I found did not explain the end of the algorithm in a way that I could understand and properly implement it.

Thank you for your help.