Publication Date:
2011-08-19
Description:
Image processing through convolution, transform coding, spatial frequency alterations, sampling, and interpolation are considered. It is postulated that convolution in one domain (real or frequency) is equivalent to multiplication in the other (frequency or real), and that the relative amplitudes of the Fourier components must be retained to reproduce any waveshape. It is suggested that all digital systems may be considered equivalent, with a frequency content approximately at the Nyquist limit, and with a Gaussian frequency response. An optimized cubic version of the interpolation continuum image is derived as a set of cubic spines. Pixel replication has been employed to enlarge the visable area of digital samples, however, suitable elimination of the extraneous high frequencies involved in the visable edges, by defocusing, is necessary to allow the underlying object represented by the data values to be seen.
Keywords:
INSTRUMENTATION AND PHOTOGRAPHY
Format:
text
Permalink
