Function TRUNCATE_FILTER

Package

Short description

Usage

    filter = TRUNCATE_FILTER(filter_f, threshold, lowpass)

Input

    filter_f (numeric): The Fourier representation of the filter.

Output

    filter (struct): The truncated representation of the filter. See descrip-
       tion for more details.

Description

    By extracting and storing only the Fourier transform coefficients whose 
    are above a certain threshold relative to the maximum value, storage
    requirements for the filters are lessened and computation is sped up
    since only non-zero coefficients are multiplied during convolution.
    The support of the Fourier transform is defined so that all coefficients
    with a magnitude above threshold*fmax are kept, where fmax is the maxi-
    mum magnitude, and so that length(filter_f) divided by the size of the
    support is a power of 2.
    The output filter contains the fields:
       filter.type (char): Fixed to 'fourier_truncated'.
       filter.N (int): The original size of the filter.
       filter.recenter (boolean): Indicates whether the Fourier transform
          should be recentered after convolution. This is always true.
       filter.start (int): The frequency index where the support of the
          Fourier transform starts.
       filter.coefft (numeric): The values of the Fourier coefficients on the
          support.

See also

• optimize_filter
• periodize_filter