FLCC documentation

  best_correlated(c::AbstractArray)

Locate the position of the element with the maximum local correlation value.

  flcc(haystack,needle)

Calculate the local (Pearson) correlation coefficients

$\mathrm{lcc}(x,y) = \frac{(x - \mu_x)^T(y - \mu_y)}{\sigma_x \sigma_y}$

between needle and all sliding windows of same size within haystack.

flcc uses the fast Fourier transform to reduce the computational complexity from O($n_H n_N$) to O(($n_H + n_N) log(n_H + n_N)$), where $n_H$ and $n_N$ are the number of elements of the haystack and the needle, respectively.

flcc supports tensors of any dimensions with real or complex entries.

Examples

Suppose you have a haystack, a tensor of reals and a needle, a smaller tensor of the same dimensionality that you are are trying to locate in the haystack. Note that the needle might be scaled and translated.

The position of the maximum element of LCC is the best match between the needle and a sliding window of the haystack

julia> using FastLocalCorrelationCoefficients

julia> haystack = rand(2^10,2^10);

julia> needle = rand(1) .* haystack[42:48, 45:50] .+ rand(1);

julia> c = flcc(haystack,needle);

julia> best_correlated(c)
CartesianIndex(42, 45)

When you need to search for many needles of the same size,

  haystack = rand(2^20);
  needle1 = rand(1) .* haystack[2:8] .+ rand(1);
  needle2 = rand(1) .* haystack[42:48] .+ rand(1);
  needle3 = rand(1) .* haystack[end-6:end] .+ rand(1);

you can preprocess the haystack to avoid redundant computations by precomputing all common information. There is no such preprocessing when using the direct method.

  precomp = flcc(haystack,size(needle1));

Then use it for much faster queries.

  best_correlated(flcc(precomp,needle1)) == 2
  best_correlated(flcc(precomp,needle2)) == 42
  best_correlated(flcc(precomp,needle3)) == 2^20-6

  lcc(haystack,needle)

Calculate the local (Pearson) correlation coefficients between a needle and a sliding window within haystack, directly.

Example

Suppose you have a haystack, a tensor of reals and a needle, a smaller tensor of the same dimensionality that you are are trying to locate in the haystack. Note that the needle might be scaled and translated.

The position of the maximum LCC is the best match between the needle and a sliding window of haystack

julia> using FastLocalCorrelationCoefficients

julia> haystack = rand(2^10,2^10);

julia> needle = rand(1) .* haystack[42:47, 45:50] .+ rand(1);

julia> c = lcc(haystack,needle);

julia> best_correlated(c)
CartesianIndex(42, 45)