vgauss -in image.v -out filtered.v -sigma 2
Three types of filtering operations are supported:
smooth This convolves the input image with a 1D
Gaussian filter in both the x and y directions
xgrad This convolves the input image with a 1D
Gaussian filter in the y direction, and a 1D
derivative-of-Gaussian filter in the x direction.
ygrad This convolves the input image with a 1D
Gaussian filter in the x direction, and a 1D
derivative-of-Gaussian filter in the y direction.
Note that the x direction is the direction of increasing
column index, and the y direction is the direction of
decreasing row index. When computing pixel values near an
image border, the region outside of the input image is
padded with border pixel values.
The input images can have pixel values of any type. In general, the output images are created with pixel values of the same type as the input images. However, if an input image has unsigned pixels (i.e., bit or ubyte) and convolution with the partial derivation of a Gaussian is required, then the input image will be converted to an image with sbyte pixels before convolution, and the output image will also have sbyte pixels.