
	This directory contains saved cantata workspaces that can
	be used for testing or demonstration purposes.
	Follow along with the explanations below, clicking on the on/off
	switches on various glyphs in order to perform the experiments.
	Note: for best results, begin each run by clicking on the "RESET" 
	button.

	example1.Z:   (restore with:  % cantata -restore example1.Z)
		   Objective: to enhance a dim image.

		   This workspace uses an image, spine.xv, which is displayed.
		   Note that the original image appears quite dim.  It is 
		   and then converted to type BYTE in order to be used as
		   to vhstr.  The vhstr program is then executed, performing 
		   a histogram stretch on the colormap of spine.xv.  
		   The image is displayed after the histogram stretch has 
		   been performed.  Notice how the spine is now much more 
		   clear, the vertibrae easier to see.

	example2.Z:   (restore with:  % cantata -restore example2.Z)
		   Objective: to use existing images to create a bizarre new
	           image.

		   This workspace uses two images to begin with, ball.xv and
		   feath.xv.   These two images are both reduced in size with
		   vshrink to make them easier to work with.  They are then
		   added using vadd. The resulting image then has a bitwise 
	 	   NOT performed on it, and the final image displayed.

	example3.Z :  (restore with:  % cantata -restore example3.Z)
		   Objective:  to introduce noise into an image, and then
		   rid the image of the noise by using median spatial filtering.

		   This workspace has an image, moon.xv, which is shrunk
	 	   down to 1/2 its normal size by vshrink.  The original,
		   smaller image is displayed.   It then has noise introduced 
		   into it by vgshot, and the image is again displayed.  
		   Note that about 25% of the image has now been corrupted 
	           with noise.  With a count loop, two iterations of vhmed 
	           are now performed, in order to do median filtering 
		   on an image using a histogram to find the median, in an 
	           attempt to reduce the noise with a window size of 3 x 3.  
		   See that each iteration takes a little more of the noise 
		   away.  A final execution of vhmed is done with a window size 
		   of 5 x 5, which does a pretty good job of filtering out
		   most of the remaining noise, and the final image is 
		   displayed by putimage.  Note that while the final image
		   has had almost all the noise removed, it is considerably
		   less accurate than the original image.  This is a side 
		   effect of the median filtering algorithm.

		   

		


