1i.gensigset(1)              GRASS GIS User's Manual             i.gensigset(1)
2
3
4

NAME

6       i.gensigset  - Generates statistics for i.smap from raster map.
7

KEYWORDS

9       imagery, classification, supervised classification, SMAP, signatures
10

SYNOPSIS

12       i.gensigset
13       i.gensigset --help
14       i.gensigset   trainingmap=name   group=name   subgroup=name  signature‐
15       file=name   [maxsig=integer]    [--overwrite]   [--help]    [--verbose]
16       [--quiet]  [--ui]
17
18   Flags:
19       --overwrite
20           Allow output files to overwrite existing files
21
22       --help
23           Print usage summary
24
25       --verbose
26           Verbose module output
27
28       --quiet
29           Quiet module output
30
31       --ui
32           Force launching GUI dialog
33
34   Parameters:
35       trainingmap=name [required]
36           Ground truth training map
37
38       group=name [required]
39           Name of input imagery group
40
41       subgroup=name [required]
42           Name of input imagery subgroup
43
44       signaturefile=name [required]
45           Name for output file containing result signatures
46
47       maxsig=integer
48           Maximum number of sub-signatures in any class
49           Default: 5
50

DESCRIPTION

52       i.gensigset  is  a  non-interactive  method  for  generating input into
53       i.smap.  It is used as the first pass in the a two-pass  classification
54       process.   It  reads a raster map layer, called the training map, which
55       has some of the pixels or regions already classified.  i.gensigset will
56       then extract spectral signatures from an image based on the classifica‐
57       tion of the pixels in the training map and make these signatures avail‐
58       able to i.smap.
59
60       The  user would then execute the GRASS program i.smap to create the fi‐
61       nal classified map.
62
63       For all raster maps used to generate signature file it  is  recommended
64       to  have  semantic  label set.  Use r.support to set semantic labels of
65       each member of the imagery group.  Signatures generated for  one  scene
66       are suitable for classification of other scenes as long as they consist
67       of same raster bands (semantic labels match). If  semantic  labels  are
68       not set, it will be possible to use obtained signature file to classify
69       only the same imagery group used for generating signatures.
70
71       An usage example can be found in i.smap documentation.
72

OPTIONS

74   Parameters
75       trainingmap=name
76           ground truth training map
77
78       This raster layer, supplied as input by the user, has some of its  pix‐
79       els  already classified, and the rest (probably most) of the pixels un‐
80       classified.  Classified means that the pixel has a non-zero  value  and
81       unclassified means that the pixel has a zero value.
82
83       This map must be prepared by the user in advance by using a combination
84       of wxGUI vector digitizer and v.to.rast, or some other  import/develop‐
85       ment  process (e.g., v.transects) to define the areas representative of
86       the classes in the image.
87
88       At present, there is no fully-interactive  tool  specifically  designed
89       for producing this layer.
90
91       group=name
92           imagery group
93
94       This  is  the name of the group that contains the band files which com‐
95       prise the image to be analyzed. The i.group command  is  used  to  con‐
96       struct groups of raster layers which comprise an image.
97
98       subgroup=name
99           subgroup containing image files
100
101       This  names  the subgroup within the group that selects a subset of the
102       bands to be analyzed. The i.group command is also used to prepare  this
103       subgroup.  The subgroup mechanism allows the user to select a subset of
104       all the band files that form an image.
105
106       signaturefile=name
107           resultant signature file
108
109       This is the resultant signature file (containing the means and  covari‐
110       ance  matrices)  for  each class in the training map that is associated
111       with the band files in the subgroup selected.
112
113       maxsig=value
114           maximum number of sub-signatures in any class
115           default: 5
116
117       The spectral signatures which are produced by this program are  "mixed"
118       signatures  (see NOTES).  Each signature contains one or more subsigna‐
119       tures (represeting subclasses).  The algorithm in this  program  starts
120       with  a maximum number of subclasses and reduces this number to a mini‐
121       mal number of subclasses which are spectrally distinct.  The  user  has
122       the option to set this starting value with this option.
123

NOTES

125       The  algorithm  in  i.gensigset determines the parameters of a spectral
126       class model known as a Gaussian mixture distribution.   The  parameters
127       are  estimated  using multispectral image data and a training map which
128       labels the class of a subset of the image pixels.   The  mixture  class
129       parameters are stored as a class signature which can be used for subse‐
130       quent segmentation (i.e., classification) of the multispectral image.
131
132       The Gaussian mixture class is a useful model because it can be used  to
133       describe  the  behavior  of  an information class which contains pixels
134       with a variety of distinct spectral characteristics.  For example, for‐
135       est, grasslands or urban areas are examples of information classes that
136       a user may wish to separate in an image.  However, each of these infor‐
137       mation  classes  may  contain  subclasses each with its own distinctive
138       spectral characteristic.  For example, a forest may contain  a  variety
139       of different tree species each with its own spectral behavior.
140
141       The objective of mixture classes is to improve segmentation performance
142       by modeling each information class as a probabilistic  mixture  with  a
143       variety  of  subclasses.  The mixture class model also removes the need
144       to perform an initial unsupervised segmentation  for  the  purposes  of
145       identifying  these  subclasses.   However, if misclassified samples are
146       used in the training process, these erroneous samples may be grouped as
147       a separate undesired subclass.  Therefore, care should be taken to pro‐
148       vided accurate training data.
149
150       This clustering algorithm estimates both the number  of  distinct  sub‐
151       classes  in  each  class, and the spectral mean and covariance for each
152       subclass.  The number of subclasses is estimated using Rissanen’s mini‐
153       mum  description  length (MDL) criteria [1].  This criteria attempts to
154       determine the number of subclasses which "best" describe the data.  The
155       approximate  maximum likelihood estimates of the mean and covariance of
156       the subclasses are computed using the expectation maximization (EM) al‐
157       gorithm [2,3].
158

WARNINGS

160       If warnings like this occur, reducing the remaining classes to 0:
161       ...
162       WARNING: Removed a singular subsignature number 1 (4 remain)
163       WARNING: Removed a singular subsignature number 1 (3 remain)
164       WARNING: Removed a singular subsignature number 1 (2 remain)
165       WARNING: Removed a singular subsignature number 1 (1 remain)
166       WARNING: Unreliable clustering. Try a smaller initial number of clusters
167       WARNING: Removed a singular subsignature number 1 (-1 remain)
168       WARNING: Unreliable clustering. Try a smaller initial number of clusters
169       Number of subclasses is 0
170       then the user should check for:
171
172           •   the  range of the input data should be between 0 and 100 or 255
173               but not between 0.0 and  1.0  (r.info  and  r.univar  show  the
174               range)
175
176           •   the  training areas need to contain a sufficient amount of pix‐
177               els
178

REFERENCES

180           •   J. Rissanen, "A Universal Prior for Integers and Estimation  by
181               Minimum Description Length," Annals of Statistics, vol. 11, no.
182               2, pp. 417-431, 1983.
183
184           •   A. Dempster, N. Laird and D. Rubin,  "Maximum  Likelihood  from
185               Incomplete Data via the EM Algorithm," J. Roy. Statist. Soc. B,
186               vol. 39, no. 1, pp. 1-38, 1977.
187
188           •   E. Redner and H. Walker, "Mixture Densities, Maximum Likelihood
189               and the EM Algorithm," SIAM Review, vol. 26, no. 2, April 1984.
190

SEE ALSO

192        r.support, i.group, i.smap, r.info, r.univar, wxGUI vector digitizer
193

AUTHORS

195       Charles Bouman, School of Electrical Engineering, Purdue University
196       Michael Shapiro, U.S.Army Construction Engineering Research Laboratory
197       Semantic label support: Maris Nartiss, University of Latvia
198

SOURCE CODE

200       Available at: i.gensigset source code (history)
201
202       Accessed: Saturday Jan 21 20:40:38 2023
203
204       Main  index | Imagery index | Topics index | Keywords index | Graphical
205       index | Full index
206
207       © 2003-2023 GRASS Development Team, GRASS GIS 8.2.1 Reference Manual
208
209
210
211GRASS 8.2.1                                                     i.gensigset(1)
Impressum