Function:     CONVOLVE

Purpose:      Convolves data with two dimensional convolution function.
              Takes care of BLANKs.

Category:     MATH

File:         convolve.c

Author:       K.G. Begeman

Use:          INTEGER CONVOLVE( CONFIE ,      Input    REAL ARRAY
                                NCONX  ,      Input    INTEGER
                                NCONY  ,      Input    INTEGER
                                DATIN  ,      Input    REAL ARRAY
                                DATOUT ,      Output   REAL ARRAY
                                NDATX  ,      Input    INTEGER
                                NDATY  )      Input    INTEGER

              CONVOLVE  Returns:
                        0 : succesfull convolution
                        1 : NDATX < NCONX
                        2 : NDATY < NCONY
                        3 : NCONX not an odd number
                        4 : NCONY not an odd number
              CONFIE    Array containing the convolution function.
                        For a convolution function c(x,y), the
                        function values are stored as follows:
                        CONFIE(K) = c(x,y) where
                        K = (y+(NCONY-1)/2)*NCONX+x+(NCONX-1)/2+1
                        So the centre of the convolution function is
                        at (NCONY-1)/2 * NCONX + (NCONX-1)/2 + 1, and
                        NCONX and NCONY must be odd. When ever a value
                        in CONFIE is equal to zero it will not be used
                        i.e. BLANKs will not propagate.
              NCONX     Number of pixels along the X axis of the
                        convolution function.
              NCONY     Number of pixels along the Y axis of the
                        convolution function.
              DATIN     Array containing the data to be convolved.
                        The size of DATIN is equalt to NDATX * NDATY.
              DATOUT    Array containing the convolved data. The
                        size of the array is equal to the size of
                        DATIN. DATOUT will contain BLANKs where the
                        convolution is not valid. This can be because
                        of BLANKs in DATIN, or because the distance
                        from the edge is less than (NCONX-1)/2 or
                        (NCONY-1)/2.
              NDATX     Number of pixels along the X axis of the
                        data to be convolved.
              NDATY     Number of pixels along the Y axis of the
                        data to be convolved.

Update:       Apr  8, 1990: KGB, Document created.