DWT uses a kernel (the wavelet), LPC models the signal using its own past samples without a pre-selected kernel. I quite like LPC ou autoregressive modelling, as I prefer to call it.
Could you please refer to this article:
http://asmp.eurasipjournals.com/content/pdf/1687-4722-2012-7.pdf
Dear @Fernando Schilndwein
Could you please explain more ? Which advantages do you thing LPC has over DWT ?
Dear @Djamel Teguig
Thanks for your interest. However, this article is related to feature extraction. I need somethings related to the computational cost, time, real time and feasibility.
Thanks.
The main advantage of not using a kernel is that you are not biasing the shape of your signal to be similar to an arbitrary, pre-selected shape. By using LPC you make only the assumption that the signal can be modelled from its past behaviour. In terms of computing, the Yule-Walker approach is quite efficient for LPC, and althought some authors question it, I found that it always gave me good results fof the signals I used (Doppler ultrasound derived blood flow velociy and heart rate variability). Best of luck.
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training