Documentation on smooth

Task: smooth
Purpose: Convolve an image (in the image domain) with a 2-D gaussian
Categories: analysis

        SMOOTH is a MIRIAD task which convolves an image by an elliptical
        gaussian or a boxcar the hard way.   The convolving Gaussian and 
        boxcar have peaks of unity.  Additional scaling is provided by 
        the keyword "scale"

Key: in
        The input image.  Wild card expansion is supported, no default.

Key: out
        The output image.  No default.

Key: type
        Specifies the type of function to convolve by.  Should
        be one of
          "gaussian"     Gaussian at arbitrary position angle 
          "boxcar"       Boxcar oriented in x and y directions. Note
                         that the full width is rounded up to be an 
        	         odd number of pixels.
        Default is gaussian.  Minimum match is active. 

Key: fwhm
        The Gaussian FWHM along the major and minor axes or the boxcar
        full widths in the x and y directions (all in arcseconds).  
        The image pixel increments are assumed to be in radians.
        No default.

Key: pa
        The position angle in degrees CCW from North of the major axis
        of the gaussian.  Not used for boxcar smoothing.
        Default is 0.0.

Key: scale
        If unset, then SMOOTH will attempt to make the units of the
        smoothed image be Jy/beam for Gaussian convolution.  If 0.0,
        then the convolution integral is scaled (multipled) by the 
        inverse of the volume of the convolving function. Otherwise, 
        the integral is scaled by "scale"

Key: options
        "nocheck"   By default, blanked input pixels do not contribute to the
           convolution sum.  If you set NOCHECK then blanked input pixels 
           are not checked for (but the output image is blanked around the 
           unconvolved edge, and wherever the input image is blanked).