Near Set Evaluation And Recognition (NEAR) System - Rough Fuzzy Image Analysis: Foundations and Methodologies - page 135

perceptual information obtained from the application of near set theory is represented by
the elementary sets (formed by the grouping of objects with similar descriptions), and the
information gained is always presented with respect to the probe functions contained inB.

7.4 Probe functions

This section describes the probe functions used in the NEAR system, and gives example
NEAR system output images processed using these probe functions.

7.4.1 Average greyscale value
Conversion from a RGB image to greyscale is accomplished using Magick++, the object-
orientated C++ API to the ImageMagick image-processing library (Magick++, 2009).
First, an RGB image is converted to greyscale using

Gr = 0.299R + 0.587G + 0.114B, (7.5)

and then the values are averaged over each subimage. An example is given in Fig. 7.5.

(7.5a) (7.5b) average greyscale over

subimages of size 5 × 5

(7.5c)

FIGURE 7.5: Example of average greyscale probe function: (a) Original image (Weber,
1999), (b) average greyscale over subimages of size 5×5, and (c) average greyscale over
subimages of size 10×10.

7.4.2 Normalized RGB
The normalized RGB values is a feature described in (Marti et al., 2001), and the formula
is given by

X
R T + G T + B T

N X = ,

where the values R T ,G T , and B T are respectively the sum of R,G,B components of the
pixels in each subimage, and X∈[R T ,G T ,B T ]. See Fig. 7.6 for an example using this
probe function. Note, these images were produces by finding the normalized value and
multiplying it by 255.

7.4.3 Shannon's entropy