r.texture(1)

1r.texture(1)                  Grass User's Manual                 r.texture(1)
2
3
4

NAME

6       r.texture  - Generate images with textural features from a raster map
7

KEYWORDS

9       raster
10

SYNOPSIS

12       r.texture
13       r.texture help
14       r.texture  [-Nqackviswxedpmno]  input=string  prefix=string  size=value
15       distance=value  [--overwrite]
16
17   Flags:
18       -N  Normalized
19
20       -q  Quiet
21
22       -a  Angular Second Moment
23
24       -c  Contrast
25
26       -k  Correlation
27
28       -v  Variance
29
30       -i  Inverse Diff Moment
31
32       -s  Sum Average
33
34       -w  Sum Variance
35
36       -x  Sum Entropy
37
38       -e  Entropy
39
40       -d  Difference Variance
41
42       -p  Difference Entropy
43
44       -m  Measure of Correlation-1
45
46       -n  Measure of Correlation-2
47
48       -o  Max Correlation Coeff
49
50       --overwrite
51
52   Parameters:
53       input=string
54           Name of the input raster map
55
56       prefix=string
57           Name of the ouput raster map
58
59       size=value
60           The size of the sliding window (odd and >= 3)
61
62       distance=value
63           The distance between two samples (>= 1)
64

DESCRIPTION

66       r.texture - Creates map raster with textural features  for  user-speci‐
67       fied raster map layer. The module calculates textural features based on
68       spatial dependence matrices at 0, 45, 90, and 135 degrees  for  a  dis‐
69       tance (default = 1).
70
71       r.texture  reads  a  GRASS  raster map as input and calculates textural
72       features based on spatial dependence matrices  for  north-south,  east-
73       west,  northwest,  and southwest directions using a side by side neigh‐
74       borhood (i.e., a distance of 1). Be sure to carefully set your  resolu‐
75       tion  (using  g.region)  before running this program, or else your com‐
76       puter could run out of memory.  Also, make sure that  your  raster  map
77       has  no more than 255 categories.  The output consists into four images
78       for each textural feature, one for every direction.
79
80       A commonly used texture model is based on the so-called grey level  co-
81       occurrence  matrix.  This matrix is a two-dimensional histogram of grey
82       levels for a pair of pixels which are  separated  by  a  fixed  spatial
83       relationship.   The matrix approximates the joint probability distribu‐
84       tion of a pair of pixels.  Several texture measures are  directly  com‐
85       puted from the grey level co-occurrence matrix.
86
87       The following are brief explanations of texture measures:
88
89                      Angular  Second Moment: This is a measure of local homo‐
90                     geneity and the opposite of Entropy.  It is high when the
91                     local window a few pixels with high values; low, when the
92                     pixels are almost equal.
93
94                      Contrast: This measure considers  the  amount  of  local
95                     variation  and  is the opposite of Homogeneity (when high
96                     pixel values concentrate along the diagonal).
97
98                      Correlation: This measure  analyses  the  linear  depen‐
99                     dency  of  grey  levels  of neighboring pixels. Typically
100                     high, when the scale of local texture is larger than  the
101                     distance.
102
103                      Entropy:  This  measure  is  high when the values of the
104                     local window have similar values.  It  is  low  when  the
105                     values  are  close to either 0 or 1 (i.e. when the pixels
106                     in the local window are uniform).
107

NOTES

109       Algorithm taken from:
110       Haralick, R.M., K. Shanmugam, and I. Dinstein. 1973. Textural  features
111       for image classification. IEEE Transactions on Systems, Man, and Cyber‐
112       netics, SMC-3(6):610-621.
113
114       The code was taken by  permission  from  pgmtexture,  part  of  PBMPLUS
115       (Copyright  1991,  Jef Poskanser and Texas Agricultural Experiment Sta‐
116       tion, employer for hire of James Darrell McCauley).
117       Man page of pgmtexture
118

BUGS

120       - The program can run incredibly slow for large raster files.
121
122       - The method for finding the  maximal  correlation  coefficient,  which
123       requires  finding the second largest eigenvalue of a matrix Q, does not
124       always converge.
125

REFERENCES

127       Haralick, R.M., K. Shanmugam, and I. Dinstein (1973). Textural features
128       for image classification. IEEE Transactions on Systems, Man, and Cyber‐
129       netics, SMC-3(6):610-621.
130
131       Bouman C. A., Shapiro M.,(March 1994).A Multiscale Random  Field  Model
132       for  Bayesian Image Segmentation, IEEE Trans. on Image Processing, vol.
133       3, no.2.
134
135       Haralick R., (May 1979). Statistical and structural approaches to  tex‐
136       ture, Proceedings of the IEEE, vol. 67, No.5, pp. 786-804
137

AUTHOR

146       G.  Antoniol  -  RCOST  (Research Centre on Software Technology - Viale
147       Traiano - 82100 Benevento.
148       C. Basco -  RCOST (Research Centre on Software Technology - Viale  Tra‐
149       iano - 82100 Benevento.
150       M. Ceccarelli - Facolt&agrave; di Scienze, Universit&agrave; del Sannio
151       Via Port&#8217;Arsa 11, Benevento.
152
153
154       Last changed: $Date: 2005/06/27 00:18:31 $
155
156       Full index
157
158
159
160GRASS 6.2.2                                                       r.texture(1)