Order of applying band-pass filtering, normalization, segmentation

Question

My purpose is to obtain several successive segments that are already band-pass filtered, centered and scaled from original signal in Matlab. I have some ways to do this

1. I apply Butter-worth and then I normalize by using X-mean(X) for zero mean, X/norm(X) for entire data. After that I will segment
2. I normalize first and use filter. At last I will segment
3. I segment first and then filter versus filter first

The reason is to examine

1. Change order of filter and normalization step will obtain same result
2. Is there different if i do same filtering for entire signal versus for each segment

I made sample example I saw that there is no difference. But I am not sure with real signal such as EEG.

Thanks for the help in advance

Answer 1

One usually wants to filter first, normalize next (since the normalization may change depending on the DC response of the filter), then segment (so only the first segment(s) incur the filter's initial transient response).