-double
-SignalFilter::integral_abscos (double u, double w)
-{
- if (fabs (u) > F_EPSILON)
- return (cos(u * w) - 1) / (u * u) + w / u * sin (u * w);
- else
- return (w * w / 2);
-}
-
-
-/* NAME
- * convolve Discrete convolution of two functions
- *
- * SYNOPSIS
- * r = convolve (f1, f2, dx, n, np, func_type)
- * double r Convolved result
- * double f1[], f2[] Functions to be convolved
- * double dx Difference between successive x values
- * int n Array index to center convolution about
- * int np Number of points in f1 array
- * int func_type EVEN or ODD or EVEN_AND_ODD function f2
- *
- * NOTES
- * f1 is the projection data, its indices range from 0 to np - 1.
- * The index for f2, the filter, ranges from -(np-1) to (np-1).
- * There are 3 ways to handle the negative vertices of f2:
- * 1. If we know f2 is an EVEN function, then f2[-n] = f2[n].
- * All filters used in reconstruction are even.
- * 2. If we know f2 is an ODD function, then f2[-n] = -f2[n]
- * 3. If f2 is both ODD AND EVEN, then we must store the value of f2
- * for negative indices. Since f2 must range from -(np-1) to (np-1),
- * if we add (np - 1) to f2's array index, then f2's index will
- * range from 0 to 2 * (np - 1), and the origin, x = 0, will be
- * stored at f2[np-1].
- */