# AutoCorrelationResampling

This package provides means to resample autocorrelation functions.

Autocorrelation function of one variable is a Laurent polynomial on ℝ in the
form

This package provides resampling of autocorrelation function in the sense that
it changes the degree of

Technically, this package works with autocorrelation function of three variables
`(x, y, z)`

becomes an array of
shape `(x, y, nz)`

where `n`

is a resampling factor.

## Why is it needed?

This package is an attempt to solve the problem of reconstruction of a porous
media when only a fraction of information about original media is available
(e.g. you have to reconstruct 3D cube from a stack of 2D slices taken along `z`

axis). If those slices are evenly sampled (i.e. you have each `n`

-th slice), you
can do the following:

- Stack the slices in 3D array. This array will have a length along
`z`

axis reduced by`n`

times. - Calculate autocorrelation function (also known as two-point function) for the reduced array.
- Upsample it by
`n`

times using this package - Reconstruct original 3D image (you can use
`PhaseRec.jl`

package for it).

## How to use?

Resampling is done by `ac_resample`

function which takes an autocorrelation
array and a resample ratio. An optional low-pass filter can be passed to
`ac_resample`

and by default is obtained with `filter_coeffs`

function. These
filters are not unique, you can play with them supplying different argument
`initial`

to `filter_coeffs`

.