I have *almost* finished a problem with symmetric extension and I am wondering if anyone here is familiar with it and has maybe gotten the parameters 100% right, as I feel I have not. The idea is this...
You can take a finite input signal and extend the start and end of it by adding a mirror image of the front and end. Then you run it through a filter bank, apply a window and decimate.
For reconstruction, you do the opposite... extend, interpolate, convolve, and window.
Anyone done this before? Mine is almost working, but I am not sure how to determine the truncation lengths and whether or not I should keep the first or 2nd. sample on decimation and interpolation. I seem to get either a small error at the endpoints or a phase shift of one sample.
-- Jason - NOSPAMbold@cae.wisc.edu Of course I always think my opinion is right! If I didn't, it wouldn't be my opinion now would it? =============================================================================