Help for MEDIAN

PURPOSE:

MEDIAN is a VICAR applications program which performs nonlinear spatial
filtering of an image based upon the local (rectangular window) median or
percentile rank of the input. Output may be in either lowpass or highpass form.

EXECUTION:

MEDIAN may be executed in the following manner:

		MEDIAN INP=A OUT=B SIZE=(SL,SS,NL,NS) PARAMS

where INP, OUT, SIZE, AND PARAMS are parameters and are explained in their
respective parameter section.

OPERATION:

MEDIAN finds the local median value of a rectangular window centered at
each pixel by accumulating  a histogram of all contained pixels. This
histogram could be interrogated for three basic statistical quantities:
mean, mode, and median. The mean is the average value in the histogram, the
mode is the most frequently occurring value in the histogram, and the median
is the value which equally divides the population of the histogram. 

The output value of MEDIAN may be modified by the inclusion of the PERCENT
keyword.  For example, PERCENT=25.0 will output the value which has 25% of 
the window pixels less than it, and 75% of the window pixels greater than it.

If HIGHPASS is specified, the final output is given as:

		OUT = IN - MEDIAN + DCLEVEL + DCTRAN*IN

BIBLIOGRAPHY:

Huang, Yang, and Tang, "A Fast Two-Dimensional Median Filtering Algorithm,"
	IEEE Trans., Vol. ASSP-27, No. 1, February 1979.
EXAMPLES:

	1) MEDIAN INP=A OUT=B NLW=5 NSW=7

		This example performs the lowpass median filter of size 5 lines
		by 7 samples.

	2) MEDIAN INP=A OUT=B 'HIGH NLW=3 NSW=3

		This example performs the highpass median filter of size
		3x3 pixels.

	3) MEDIAN INP=A OUT=B PERCENT=20 NLW=7 NSW=9

		This example outputs a value corresponding to the 20% level
 		of a histogram generated by a 7 line by 9 sample filter. The
		output is then the 12th lowest value in the local window
		rather than the 31st lowest value (50% level).
HISTORY:

ORIGINALLY WRITTEN BY: W. D. Benton, 27 November 1978
REWRITTEN WITH FASTER ALGORITHM BY:  H. J. Frieden,  22 July 1980
CONVERTED TO VAX BY:  Helen De Rueda,  30 Nov. 1983
REWRITTEN TO PERMIT NON-BYTE DATA: Ron Alley, 4 Dec 1991
COGNIZANT PROGRAMMER:  Ron Alley

PARAMETERS:


INP

input dataset.

OUT

output dataset.

SIZE

VICAR size field, (SL,SS,NL,NS).

SL

starting line to be output

SS

first sample to be output

NL

number of lines

NS

Number of samples

NLW

Size of filter in lines

NSW

Size of filter in samples

HIGHPASS

KEYWORD-Highpass option.

DCLEVEL

Offset to be added to highpass output.

PERCENT

Output value percentile

DCTRAN

DCTRAN*Local Median is added to the highpass output.

See Examples:


Cognizant Programmer: