This program will apply a horizontal or vertical derivative to a cube.
This will highlight edges in the image created by groups of contrasting
DN values. The horizontal derivative utilizes a two sample by one line
boxcar convolution for its solution, and will highlight horizontal linear
changes. The vertical derivative utilizes a one sample by two line boxcar
convolution for the output, and will highlight vertical linear changes.
Please refer to the following 2 by 2 pixel example image:

In this image, the horizontal derivative for A would be:

A_NewDN = A - B

The vertical derivative for A would be:

A_NewDN = A - C

NOTE: If this program needs to do calculations where special pixels
are involved, it will simply output NULL. If this causes undesireable
results, you should run the image through the stretch program to change
all special pixels to valid data. For example, if you have a number of
HRS pixels, you might want to stretch those to 255 prior to running the
deriv filter.