GNG

Description

Adds random Gaussian noise to an existing image.

Parameters

gng(file, dboc, dbow, mean, stdv, seed)

Name	Type	Caption	Length	Value range
FILE *	str	Input file name	1 -
DBOC *	List[int]	Random Noise Layer	1 -
DBOW	List[int]	Raster output window	0 - 4	0 -
MEAN *	List[float]	Mean value	1 - 1
STDV *	List[float]	Standard deviation	1 - 1	0 -
SEED	List[int]	Random number seed (odd)	0 - 1	101 - 32767 Default: 1357

* Required parameter

Parameter descriptions

FILE

Specifies the name of the input file that contains the layer(s) to be processed.

DBOC

Specifies the output channel(s) or layer(s) to receive the output results.

Duplicate channels are not allowed.

DBOW

Specifies the raster window (Xoffset, Yoffset, Xsize, Ysize) to be processed. If DBOW is not specified, the entire layer is output by default. Xoffset and Yoffset define the upper-left starting pixel coordinates of the window. Xsize is the number of pixels that define the window width. Ysize is the number of lines that define the window height.

MEAN

Specifies the average gray-level value of Gaussian noise.

STDV

SEED

Specifies the random number seed.

This value must be an odd integer, between 101 and 32767.

Details

GNG generates Gaussian noise and adds it to a specified set of output image channels (DBOC) in the specified file. GNG is used to test the robustness of various image analysis algorithms, such as classifiers, filters, and transforms, in the presence of noise. The Gaussian noise has a specified mean (MEAN) and standard deviation (STDV), defined in gray-level units. When adding noise to existing image data, the mean is typically set to 0.

Gaussian noise can be added over a specified output window (DBOW), if desired.

A random number seed (SEED) must be specified and must be an odd number. This allows the generation of different Gaussian images (based on different seeds) having the same mean and standard deviation.

Example

Add random Gaussian noise (with a mean of 0 and 34% of the additive noise in the gray level range 0-3) to an existing image channel (1-4) from the file 'irvine.pix' in window 16,16,64,64.

from pci.gng import gng

file     = 'irvine.pix'   #Input file name
dboc     = [1,2,3,4]    #add noise to channels 1-4
mean     = [0.0]          #noise has a mean 0.0
stdv     = [3.0]          #1 34% of the additive noise has gray level of 0-3
seed     = [111]          #random number seed 111 used to generate noise
dbow     = []

gng(file, dboc, dbow, mean, stdv, seed)