public static class FastGaussCalculator.Parameters
extends Object

Collects all coefficients required to carry on the filtering, see eq. (35) in the paper. If I(x) is your input array at offset x, and O(x) is supposed to be the filtered output, one should do:

tmp = I(x-N-1) + I(x+N-1); O(x) = nk_2[0] * tmp - dk_1[0] * yk(x-1) - yk(x-2); O(x) += nk_2[1] * tmp - dk_1[1] * yk(x-1) - yk(x-2); O(x) += nk_2[2] * tmp - dk_1[2] * yk(x-1) - yk(x-2); if M == 4: O(x) += nk_2[3] * tmp - dk_1[3] * yk(x-1) - yk(x-2);

This way 2*M multiplications and 3*M additions are made per one 'x'. Note that for-loop is not welcome as it involves comparison-test and additional addition per one 'x'.

The class also stores Sigma for which the coefficients are valid (as long as user is not changing values arbitrarily...).