rboxfilter

From DavinciWiki

Contents: Description, Arguments, Usage, Examples, Images, Sub-Functions, Related Functions, Source

Description

Perform a running boxfilter on every xy plane of the data (band independent).

rboxfilter() performs an individual boxfilter on every sub-section of the array defined by an x and y dimension. It then uses an overlap of half of the defined size to blend it with the next section of the data. This is done so there is no visible seam and the stretch is essentially "running" or changing over the whole data set. This process helps reduce the huge memory dependance of boxfilter allowing much larger images (>5x larger) to be processed using boxfilter(). While it is not as fast as boxfilter, it is still much faster than any of the convolve functions.

Changing the x and y sizes will ultimately determine how local or regional features affect the stretch.

Arguments and Return Values

Arguments: Accepts any numeric array

Return Value: A float array

Usage

Syntax: rboxfilter(data = VAL [, ignore = FLOAT] [, ysize = INT] [, xsize = INT] [, ksize = INT] [, verbose = BOOL])

data - Any numeric array

ignore - Ignore value of the black space. Default of -32768.

ysize - Any integer for beaking up the data in the y direction. Default of 1000.

xsize - Any integer for beaking up the data in the x direction. Default of 4000.

ksize - boxfilter kernel size (currently only accepts square kernels). Default of 25.

verbose - print the segment of the image that is being manipulated. Default is 0.

Examples

dv> a
345x600x1 array of byte, bsq format [207,000 bytes]

The two methods below are equivalent and yeild the same results, but the second can be used for more memory efficient processing.

dv> b=rboxfilter(a, xsize=50, ysize=100, ksize=21, ignore=0)
345x600x1 array of float, bsq format [828,000 bytes]

dv> b=rboxfilter(data=a, xsize=50, ysize=100, ksize=21, ignore=0)
345x600x1 array of float, bsq format [828,000 bytes]

Images